Service

Models for creating video, Text-to-video: These models convert a text prompt into a video from scratch and Image-to-video: These models use a still image as a base and a text prompt to control its motion

Flek lets you select only the areas you want to adjust, enter a prompt, and receive the new image at the same high resolution—ready for use on any media within 10 seconds.

FLUX 3 Video is a highly versatile, multi-modal model that supports text-to-video, image-to-video, and keyframe inputs, extending to advanced features like agentic chaining and video continuation. It allows users to generate up to 20-second clips featuring multiple scenes in a single take. Beyond standard cinematic styles, it offers broad stylistic diversity, native audio integration, precise text and typography rendering, and robust multilingual dialogue capabilities.

Image generation using Flux LoRA is the process of creating images by combining a base model (Flux), which excels at high-quality image generation, with a supplementary model (LoRA) that contains deep, specialized data. This combination helps us achieve images with specific, detailed characteristics.

A state-of-the-art image generation model from OpenAI that can create images from text or precisely edit existing ones based on instructions. It supports both generating new visuals and refining images while preserving key details as intended. When reference images are provided, the system automatically processes them at high fidelity—no extra settings required. You can use a single image for editing or multiple images to seamlessly combine styles and elements into one cohesive result.

Grok Imagine Image is xAI’s text-to-image model that generates images quickly (around 4 seconds per image) and supports a wide range of styles, including photorealistic, anime, oil painting, and abstract art. It excels at rendering clear, readable text within images—making it ideal for graphic design—and also allows image editing by uploading a source image and describing the desired changes.

Turn static images into cinematic videos with synced audio using xAI’s model. Upload any image and it’s brought to life with realistic motion, interactions, and matching sound—also supports text-to-video (T2V) generation.

Upscora is  AI image upscaler that boosts resolution and enhances quality by generating new high-detail pixels using a Multidiffusion Upscaler, allowing creators to refine images with prompts, Denoise, CFG, and other settings for print-ready, media-grade, or stock-photo-level results in a single streamlined step.

Varo Koneko for Adobe Photoshop AI image-editing plugin for Photoshop — select an area, describe the edit, and the result is placed as a new layer. Powered by the Varoriya API. ⬇ Download (.ccx) Get API Key Features Inpaint by selection — select an area (multiple / irregular / lasso) and the result is placed as a non-destructive Smart Object clipped with a soft, feathered mask. Whole-image edit — with no selection, the result fills the full canvas width and the height follows the generated aspect (matches the source, never stretched). Reference images — attach up to 2 reference images (click to add) alongside the canvas edit. Multiple models — Nanobanana Pro, SeeDream 5 Pro, GPT Image 2, each with quality tiers. Native Photoshop (Spectrum) UI with one-click Generate. Requirements Photoshop 24.0.0 or newer (UXP) — Windows or macOS. Adobe Creative Cloud desktop app (installs the plugin). A Varoriya API key (varo_sk_…). Install Download varo-koneko.ccx from the button above. Double-click the .ccx file — Creative Cloud installs it automatically (Windows & Mac). Open Photoshop → Plugins menu → Varo Koneko to open the panel. In the panel, open the menu (☰) → API Key, paste your API key, and Save. Usage Open an image (optionally select an area to inpaint). (Optional) Add up to 2 reference images — click the Reference box to pick files. Choose a model + quality, then type your edit prompt. Click Generate — the result is placed as a new Smart Object layer. Note: Each generation deducts coins based on the model/quality; failed jobs are auto-refunded. Most image jobs finish in seconds (the panel waits up to ~5 minutes). 📌 Follow the latest version & updates on GitHub Releases.

Nano Banana 2 and Pro deliver rapid processing for high-volume tasks and meticulous detail for professional production. They natively generate 4K images without external upscaling. These models support up to 14 reference images and seamlessly integrate PDF reading, real-time web searches, and live internet image retrieval for powerful, flawless results.

It is a model designed for editing images using natural language prompts. It is fast, precise, and capable of preserving the fine details of people and objects.

OmniHuman 2.0 is an AI video generator that transforms a single static image and an audio track into a highly realistic, expressive motion avatar. The system flawlessly interprets the audio to synchronize lip movements, facial expressions, and body language with the speech's rhythm, tone, and meaning. Whether generating cinematic, emotionally nuanced acting or soulful singing performances across any genre, the results feel incredibly natural and human. Furthermore, it supports optional text prompts, allowing you to precisely control camera movements, specific actions, and object generation for complete creative refinement in a single workflow. Read More Prompt Guide

This mode is designed for repairing old or damaged photographs. It can remove scratches, fix cracks, enhance details, and colorize black-and-white images to bring your memories back to life with vibrant clarity. Perfect for vintage portraits and historical family photos.

SeeDance  1.5 PRO A multimodal generative video foundation model with built-in V2A native generation support for coordinated audio and video output. It achieves millisecond-level synchronization between visual frames and sound (ambient, effects, dialogue, music), high semantic prompt fidelity, cinematic motion rendering, and efficient video creation directly from text and image inputs

Seedance 2.0 is a professional multi-mode video generation model by ByteDance. It supports video creation from reference media (images, videos, and audio), video editing, and video extension. Control style, camera angles, and character consistency with precision. Cinematic quality with stable, realistic output. Supports complex scenes, professional camera work, multi-shot storytelling, and multiple languages. Perfect for advertising, film, and social media marketing.

Seedance 2.5 is now on Varoriya! Experience 30-second extended storytelling, expanded multimodal references, and total prompt freedom with unrestricted capabilities.

Seedream 5.0 is ByteDance's high-end Generative AI image creation model, now available in two tiers: Lite and Pro. It stands out with its Visual Reasoning system — "thinking before drawing" — enabling it to understand complex logic and physics more accurately than typical models. It can pull real-time information from the internet to create images that stay current and relevant to today's events, while excellently maintaining character and style consistency across generations. Seedream 5.0 Pro takes it a step further for professional workflows. With deep spatial grounding and regional understanding, it enables pixel-precise, location-targeted editing, and can systematically organize complex, high-density infographics. It natively supports input and generation in over 10 languages worldwide (including Thai), and delivers more realistic lighting, shadows, and textures — built for professional design and creative work.

The latest text-to-video model from OpenAI, capable of generating cinematic-quality motion with realistic physics, natural scene continuity, and synchronized sound.It supports prompts to define scenes, voices and dialogues.

SpyShot reveals multiple hidden perspectives from a single image you upload, as if you could walk around the scene and observe it from different positions and unique angles that were never visible before.

Tweening is a video generation mode using AI that creates a smooth, realistic, and naturally continuous animation between two still images — the starting frame and the final frame — by automatically generating the in-between frames

text-to-speech system that converts written text into natural-sounding speech with customizable voices, emotions, pacing, tone, and languages, making it ideal for video narration, online content, and character voice creation without needing to record audio

WAN is a model trained on motion-based datasets. It can generate 2-megapixel images in just a few seconds

Faster, smaller, more cost-efficient, and environmentally friendly. High-performance image-to-video generation.

Photorealistic, Cost-effective, Fast & Highly Efficient

Varo Boxes  is an AI-powered image editing feature that grants users exact control over the size and placement of elements by drawing bounding boxes directly onto an image. Within these custom boundaries, users can input specific text prompts for targeted modifications or type @ to seamlessly insert uploaded reference products exactly where desired. This tool eliminates the unpredictability of standard AI generation, enabling users to flawlessly execute multiple localized edits—such as adjusting eye color, changing lip texture, and placing a specific product—in a single generation step without confusing the AI or altering the rest of the image.

MiniMax H3 is an all-around AI video generation model capable of seamlessly integrating text, images, audio, and video. It allows users to freely mix complex instructions in one go, delivering video outputs up to 15 seconds long at up to 2K resolution. Moreover, its operating cost is many times cheaper than typical models on the market

Imagen 4 from Google is part of the Generative AI category for text-to-image creation. Its key strengths lie in accuracy, realism, and ease of use, allowing users to generate professional-quality images without needing any technical expertise.

Transforms your uploaded image into an IMAX-style panoramic scene. Using AI, it analyzes the original image and expands the world in every direction, generating a surrounding environment that feels natural and cohesive—with realistic lighting, shadows, textures, and atmosphere.

Seed Audio 1.0. that can handle music, sound effects, and dialogue from references for consistent voices. Text to Background Ambient, Text to Music, Text to Dialogue and SFX, Audio Reference / Voice Cloning

Varo MCP Use Varoriya Services Directly via AI Agents Varo MCP is a connector that enables AI Agents like Claude, VS Code Copilot, Cursor, and ChatGPT to use Varoriya's content generation services directly through chat — you just type what you want, and the AI handles calling the models for you. No need to open a web browser or copy-paste. Services available via Varo MCP: Generate Images — z-image, Grok, Nano Banana Pro, Seedream 5 Pro Generate Videos — Seedance 2.0, WAN, Grok Video (with audio) Generate Audio — Music, ambiance, SFX, voiceover (Seed Audio) Blockchain Registration — Certify your work on the Varoriya Registry (Polygon) Coin Management — Check VaroCoin balance, Top up via PromptPay QR Varo MCP operates via the MCP (Model Context Protocol), an open protocol that allows AI Agents to integrate with external tools — Varo MCP is Varoriya's implementation, instantly giving access to our services for any MCP-compatible AI. Installing Varo MCP Varo MCP is provided via a Remote Server at https://mcp.varoriya.com — no additional installation required. Just add this URL to your client and log in with your Varoriya account via OAuth (no API Key needed). Claude Step-by-Step Installation on Claude Desktop Claude Desktop is a popular AI Assistant — once Varo MCP is installed, you can generate images, videos, and audio directly through your Claude chat. Open Claude Desktop Go to Settings (⚙️ bottom left or Ctrl+,) Select the Connectors tab Click Add custom connector Enter URL: https://mcp.varoriya.com Click Connect → An OAuth window will open for you to log in with your Varoriya account Once connected successfully, you will see varo in the Connectors list with a Connected ✓ status — It is recommended select Always allow so Claude can call the tool instantly without asking every time. After installing, try typing one of the examples below → Chat Prompt Examples Tip: If your Claude Desktop does not have a Connectors tab yet, update it to the latest version, or view the stdio bridge installation guide at dev.varoriya.com/mcp Chat Prompt Examples Once Varo MCP is installed, you can instruct the AI immediately. The AI will automatically call the relevant tools — you don't need to specify the tool name yourself. Generate Images "Generate a 16:9 picture of a sea of mist in the morning using z-image" "Draw a 3D cartoon style robot watering plants using Grok" "Create an image of cosmetic products on a marble background in 1K size using Seedream 5 Pro" "Edit this picture to make the whole image blue using Nano Banana Pro" (attach original image) "Modify this photo by changing the shirt to a formal suit with a plain background, keep the original face strictly unchanged" (attach original image) "Restore this old torn photo of my grandmother, fix the tears, keep the original face strictly unchanged using gptimage" (attach original image) Generate Videos "Make a 5-second video of a waterfall in a tropical rainforest with sunlight shining through leaves using Seedance 2.0" "Create an 8-second video from this picture, make the character turn to smile and wave using WAN" "Make a 4-second cinematic video of a cat walking on a fence using Grok Video" "Use this photo to make a video with Grok Video, making them say 'Hello' in Thai" (attach original image) Generate Audio "Create a beautiful 60-second Lo-fi piano instrumental for studying" "Generate 30 seconds of tropical rainforest nature sounds with birdsong and flowing water" "Create a Thai female voiceover with a warm tone saying 'Welcome to Varoriya'" Check Balance / Top Up "Please check my remaining coin balance" "Top up 300 Baht" "I want to top up coins, what packages are available?" View Models and Pricing "What models are available to use right now?" "Check the price for 2K quality z-image" "How many coins does generating an image with z-image cost?" "Which video generation model is the cheapest?" Blockchain Registration "Check the registration price for this image on the Blockchain" "Register this image on the Varoriya Registry" "This picture is amazing, register it on the blockchain" Coin Management Pre-deduct on submit — If a job fails, coins are automatically refunded Check remaining balance: varo_balance Top up coins: varo_topup — Get a PromptPay QR code to scan and pay Request an API Key for Direct API usage: varoriya.com/account/key View Generation History Go to varoriya.com/api-usage — displays your entire generation history via MCP and API, categorized by app (MCP: VS Code, MCP: Cursor, MCP: Claude, API direct) Frequently Asked Questions (FAQ) Do I need an API Key? No — Varo MCP uses OAuth to log in directly with your Varoriya account. No need to copy-paste any keys. If you want to use an API Key (e.g., for CI/CD or automated systems), you can request one at varoriya.com/account/key and add it as a header: Authorization: Bearer varo_gk_... How long are generated files kept? Output files (images/videos/audio) are kept for 24 hours on the VARORIYA CDN — which is sufficient time for downloading. If you want permanent storage, register them on the Varoriya Registry. Which Clients are supported? Client Remote HTTP stdio bridge Deeplink Notes Claude Desktop ✅ ✅ ❌ Connector is recommended (Settings → Connectors) VS Code ✅ ✅ ✅ One-click install Cursor ✅ ✅ ✅ One-click install ChatGPT ✅ ❌ ❌ Business / Enterprise / Edu plans only (beta) Claude Code (CLI) ✅ ❌ ❌ claude mcp add varo --url ... Which APIs does MCP use under the hood? Varo MCP calls the Varo Generate API (/v1/generate/image, /v1/generate/video, /v1/generate/audio) — view the full documentation at the API Reference Related Links Link Description dev.varoriya.com Developer Hub — Tools for developers dev.varoriya.com/mcp MCP Setup Page (One-click install) dev.varoriya.com/reference Generate API Reference (OpenAPI/Scalar) dev.varoriya.com/playground Playground — Try the Generate API in your browser api.varoriya.com REST API Endpoint mcp.varoriya.com MCP Server Endpoint varoriya.com/account/key Request an API Key varoriya.com/api-usage View generation history via API/MCP github.com/varoriya/varo-mcp Source Code (GitHub)

Create 360-degree images with AI from either existing images or text prompts, allowing viewers to look around in every direction as if they were standing at that location. This is ideal for wide-angle scenes, locations, and landscapes. If a reference image includes a person, it is recommended to use a full-body image so the system can generate the surrounding environment more naturally. The system supports multiple reference images, enabling the AI to use them as context for generating a complete 360° environment. The final result can also be posted and viewed directly as a 360° image on Facebook

VaroChrome 2.0: The All-in-One AI Extension to Elevate Generative AI Workflows Overview Working with AI tools today often involves complex steps and switching between multiple platforms—from finding references and adjusting prompts to generating images and uploading them for video creation. VaroChrome 2.0 simplifies this process, instantly transforming text or images into still images, videos, and voiceovers. It seamlessly connects prompt creation with precise post-processing all in one place, without leaving the current web page. Key Features Seamless Integration: Connects directly with ChatGPT, Gemini, and Pinterest. The VaroChrome icon automatically appears under images. Simply click to configure, edit, or generate videos instantly. Universal Access: For other websites, simply "right-click" on any image and select the VaroChrome menu. The extension panel will open, ready for configuration and generation. Top-Tier AI Models: Image Models: Seedream 5 Pro, Nano Banana Pro, GPT-Image 2.0, Grok Image, Z-Image Video Models: Seedance 2.0, WAN 2.2, H3, Flux 3, including Grok Video with Thai speech support Audio Models: Seed Audio in 19 languages  for speech, music, and sound effects generation How to Use Install: Download and install VaroChrome on the browser. Log in: Connect a Varoriya account (one-click) or provide a Generation API Key. Select Reference: Open any desired website. Activate: Right-click on an image → select "VaroChrome Panel" (or click the VaroChrome icon under images on supported sites). Create: Select a model, adjust settings as needed, and click Generate. Download: Save the results for immediate use. Privacy Policy Highest priority on user data security: Local Processing: All data is handled locally on the device until a media generation command is sent. Auto-Deletion: Media files generated by the system are automatically deleted from the server within 24 hours. Download VaroChrome 2.0: Chrome Web Store

vCube Video Upscaler is ByteDance’s AI-powered video enhancement technology that upscales low-resolution videos (e.g., 480p/720p to 1080p or 4K) while restoring lost details and improving overall visual quality. It offers optimized enhancement presets for five key scenarios: general content, UGC short videos, AIGC content, short dramas, and classic film restoration. By intelligently detecting issues such as compression artifacts, noise, and color distortion, vCube applies adaptive enhancement to deliver sharper, cleaner, and more vibrant videos—even at the same bitrate.

Flux Multi-LoRA is an extension of the Flux and LoRA concept, which involves using multiple LoRAs simultaneously within a single prompt.

FLUX is a text-to-image AI model developed by BFL. The model was created as a tool to help users quickly and efficiently generate high-quality images

Image generation using Flux LoRA is the process of creating images by combining a base model (Flux), which excels at high-quality image generation, with a supplementary model (LoRA) that contains deep, specialized data. This combination helps us achieve images with specific, detailed characteristics.

A state-of-the-art image generation model from OpenAI that can create images from text or precisely edit existing ones based on instructions. It supports both generating new visuals and refining images while preserving key details as intended. When reference images are provided, the system automatically processes them at high fidelity—no extra settings required. You can use a single image for editing or multiple images to seamlessly combine styles and elements into one cohesive result.

Grok Imagine Image is xAI’s text-to-image model that generates images quickly (around 4 seconds per image) and supports a wide range of styles, including photorealistic, anime, oil painting, and abstract art. It excels at rendering clear, readable text within images—making it ideal for graphic design—and also allows image editing by uploading a source image and describing the desired changes.

Grok Imagine Image is xAI’s text-to-image model that generates images quickly  and supports a wide range of styles, including photorealistic, anime, oil painting, and abstract art. It excels at rendering clear, readable text within images—making it ideal for graphic design—and also allows image editing by uploading a source image and describing the desired changes.

Nano Banana 2 and Pro deliver rapid processing for high-volume tasks and meticulous detail for professional production. They natively generate 4K images without external upscaling. These models support up to 14 reference images and seamlessly integrate PDF reading, real-time web searches, and live internet image retrieval for powerful, flawless results.

Nano Banana 2 deliver rapid processing for high-volume tasks and meticulous detail for professional production. They natively generate 4K images without external upscaling. These models support up to 14 reference images and seamlessly integrate PDF reading, real-time web searches, and live internet image retrieval for powerful, flawless results.

An AI model from ByteDance (TikTok) that unifies text-to-image generation and image editing in one tool. It supports multiple image references, maintains accurate style and character consistency, runs fast, and delivers high-quality results up to 4K resolution.

An AI model from ByteDance (TikTok) that unifies text-to-image generation and image editing in one tool. It supports multiple image references, maintains accurate style and character consistency, runs fast, and delivers high-quality results up to 4K resolution.

Seedream 5.0 is ByteDance's high-end Generative AI image creation model, now available in two tiers: Lite and Pro. It stands out with its Visual Reasoning system — "thinking before drawing" — enabling it to understand complex logic and physics more accurately than typical models. It can pull real-time information from the internet to create images that stay current and relevant to today's events, while excellently maintaining character and style consistency across generations. Seedream 5.0 Pro takes it a step further for professional workflows. With deep spatial grounding and regional understanding, it enables pixel-precise, location-targeted editing, and can systematically organize complex, high-density infographics. It natively supports input and generation in over 10 languages worldwide (including Thai), and delivers more realistic lighting, shadows, and textures — built for professional design and creative work.

A technique/model of AI used to transfer the style of one image onto another while preserving the original structure and main content of the image.

WAN is a model trained on motion-based datasets. It can generate 2-megapixel images in just a few seconds

Photorealistic, Cost-effective, Fast & Highly Efficient

Imagen 4 from Google is part of the Generative AI category for text-to-image creation. Its key strengths lie in accuracy, realism, and ease of use, allowing users to generate professional-quality images without needing any technical expertise.

Flux Hyper uses an 8-step process, generating 2 images per run. The average generation speed is about 2–3 seconds per image.

open-source multimodal image generation and editing model from Alibaba/Qwen, offering flexible stylistic control and high-quality visual outputs

Recraft V3 (code-named red_panda) is a text-to-image model capable of generating long texts and images in a wide range of styles

Stable Diffusion 3.5 Large is an image generation model from Stability AI that uses Rectified Flow for more efficient noise sampling, enabling better prompt understanding, improved text rendering, and more reliable multi-person images, hand details, and font expressions

SDXL-Lightning is a high-speed image generation model that produces SDXL-level images in just four inference steps, enabling near real-time performance for fast creative workflows.

A text-to-image model developed by ByteDance, Seedream 3.0 stands out with its ability to generate native 2K resolution images quickly and accurately, with strong alignment between visuals, text, and composition. It supports diverse fonts and styles, making it ideal for professional-grade posters, cinematic scenes, and photorealistic portraits

Flek lets you select only the areas you want to adjust, enter a prompt, and receive the new image at the same high resolution—ready for use on any media within 10 seconds.

An AI tool/model for Inpainting or Generative Fill, used to seamlessly complete missing parts of an image or modify specific areas while preserving natural context. Simply paint over any region of the image, then describe in text what you want that area to become, and the system will generate realistic results consistent with the original scene.

A state-of-the-art image generation model from OpenAI that can create images from text or precisely edit existing ones based on instructions. It supports both generating new visuals and refining images while preserving key details as intended. When reference images are provided, the system automatically processes them at high fidelity—no extra settings required. You can use a single image for editing or multiple images to seamlessly combine styles and elements into one cohesive result.

Grok Imagine Image is xAI’s text-to-image model that generates images quickly (around 4 seconds per image) and supports a wide range of styles, including photorealistic, anime, oil painting, and abstract art. It excels at rendering clear, readable text within images—making it ideal for graphic design—and also allows image editing by uploading a source image and describing the desired changes.

Grok Imagine Image is xAI’s text-to-image model that generates images quickly  and supports a wide range of styles, including photorealistic, anime, oil painting, and abstract art. It excels at rendering clear, readable text within images—making it ideal for graphic design—and also allows image editing by uploading a source image and describing the desired changes.

Image Combination Mode is a feature that allows you to merge two images into one and flexibly modify different parts of the image using text prompts. It leverages AI technology that understands the context of the images and the user's instructions

LoRA Paint is the technique of using an in-painting process together with a desired LoRA Model, which allows us to 'draw' or 'paint' the style and details from the LoRA onto any part of the image as desired

Nano Banana 2 and Pro deliver rapid processing for high-volume tasks and meticulous detail for professional production. They natively generate 4K images without external upscaling. These models support up to 14 reference images and seamlessly integrate PDF reading, real-time web searches, and live internet image retrieval for powerful, flawless results.

Nano Banana 2 deliver rapid processing for high-volume tasks and meticulous detail for professional production. They natively generate 4K images without external upscaling. These models support up to 14 reference images and seamlessly integrate PDF reading, real-time web searches, and live internet image retrieval for powerful, flawless results.

It is a model designed for editing images using natural language prompts. It is fast, precise, and capable of preserving the fine details of people and objects.

Designed for editing image with prompt. enables precise local edits without disrupting overall composition, maintains character consistency, references visual styles from source images, and delivers fast, high-performance results

An AI model from ByteDance (TikTok) that unifies text-to-image generation and image editing in one tool. It supports multiple image references, maintains accurate style and character consistency, runs fast, and delivers high-quality results up to 4K resolution.

An AI model from ByteDance (TikTok) that unifies text-to-image generation and image editing in one tool. It supports multiple image references, maintains accurate style and character consistency, runs fast, and delivers high-quality results up to 4K resolution.

Seedream 5.0 is ByteDance's high-end Generative AI image creation model, now available in two tiers: Lite and Pro. It stands out with its Visual Reasoning system — "thinking before drawing" — enabling it to understand complex logic and physics more accurately than typical models. It can pull real-time information from the internet to create images that stay current and relevant to today's events, while excellently maintaining character and style consistency across generations. Seedream 5.0 Pro takes it a step further for professional workflows. With deep spatial grounding and regional understanding, it enables pixel-precise, location-targeted editing, and can systematically organize complex, high-density infographics. It natively supports input and generation in over 10 languages worldwide (including Thai), and delivers more realistic lighting, shadows, and textures — built for professional design and creative work.

Varoriya Light & BG Generator is an AI-powered tool that allows users to adjust the light direction, lighting conditions, background, and even add custom prompts to their images. The AI then generates a brand-new image within just a few seconds

a deep-learning–powered tool that swaps faces between images with high accuracy by detecting facial features and blending them seamlessly into the target media.

Upscora is  AI image upscaler that boosts resolution and enhances quality by generating new high-detail pixels using a Multidiffusion Upscaler, allowing creators to refine images with prompts, Denoise, CFG, and other settings for print-ready, media-grade, or stock-photo-level results in a single streamlined step.

This mode is designed for repairing old or damaged photographs. It can remove scratches, fix cracks, enhance details, and colorize black-and-white images to bring your memories back to life with vibrant clarity. Perfect for vintage portraits and historical family photos.

An AI model for upscaling images to higher resolution and sharper quality, designed to realistically enhance low-quality, blurry, or pixelated images

Video Upscaler reconstructs detail and sharpens your video — take a 480p, 720p, or 1080p clip and upscale it anywhere from 720p up to 4K, with frame rates from 24 to 120 fps. Just upload a clip, select the range you want, choose a resolution and frame rate, and let the AI handle the rest.

An AI-powered sharpening tool that enhances image clarity not only by refining existing pixel edges but also by generating new, high-detail pixels for a sharper, more defined result

vCube Video Upscaler is ByteDance’s AI-powered video enhancement technology that upscales low-resolution videos (e.g., 480p/720p to 1080p or 4K) while restoring lost details and improving overall visual quality. It offers optimized enhancement presets for five key scenarios: general content, UGC short videos, AIGC content, short dramas, and classic film restoration. By intelligently detecting issues such as compression artifacts, noise, and color distortion, vCube applies adaptive enhancement to deliver sharper, cleaner, and more vibrant videos—even at the same bitrate.

Models for creating video, Text-to-video: These models convert a text prompt into a video from scratch and Image-to-video: These models use a still image as a base and a text prompt to control its motion

a video generator that extracts motion and voice from a source video, then applies them onto an uploaded image — allowing the image to move, speak, and animate just like the original. The output video’s aspect ratio is based on the proportions of the uploaded image.

AI model that transforms a single still image into a fully animated character by combining it with a driving video. Using just one reference image and a motion source video, the system accurately transfers facial expressions, gestures, and full-body movements onto the character. It works with real humans, cartoons, animals, and fantasy characters, supporting both portrait and full-body shots. Most importantly, it preserves the character’s identity and visual details while delivering precise and natural motion at the same time.

FLUX 3 Video is a highly versatile, multi-modal model that supports text-to-video, image-to-video, and keyframe inputs, extending to advanced features like agentic chaining and video continuation. It allows users to generate up to 20-second clips featuring multiple scenes in a single take. Beyond standard cinematic styles, it offers broad stylistic diversity, native audio integration, precise text and typography rendering, and robust multilingual dialogue capabilities.

Turn static images into cinematic videos with synced audio using xAI’s model. Upload any image and it’s brought to life with realistic motion, interactions, and matching sound—also supports text-to-video (T2V) generation.

A lipsync service built on Seedance 2.0, with UX/UI designed for singing in any language. Handles camera movement through descriptive prompts — along with multiple reference image types.

A highly dynamic and realistic video generation model that creates videos from a single image using text prompts combined with voice or music input. It supports native 1080p video output, enables accurate lip-syncing based on the provided audio, supports multi-character performances, and allows precise control over character gestures and camera movement.

OmniHuman 2.0 is an AI video generator that transforms a single static image and an audio track into a highly realistic, expressive motion avatar. The system flawlessly interprets the audio to synchronize lip movements, facial expressions, and body language with the speech's rhythm, tone, and meaning. Whether generating cinematic, emotionally nuanced acting or soulful singing performances across any genre, the results feel incredibly natural and human. Furthermore, it supports optional text prompts, allowing you to precisely control camera movements, specific actions, and object generation for complete creative refinement in a single workflow. Read More Prompt Guide

This is an AI tool for generating videos. It replaces the character in an original video with a character from an image you upload, which then mimics the movements and audio of the source video

SeeDance  1.5 PRO A multimodal generative video foundation model with built-in V2A native generation support for coordinated audio and video output. It achieves millisecond-level synchronization between visual frames and sound (ambient, effects, dialogue, music), high semantic prompt fidelity, cinematic motion rendering, and efficient video creation directly from text and image inputs

Seedance 2.0 is a professional multi-mode video generation model by ByteDance. It supports video creation from reference media (images, videos, and audio), video editing, and video extension. Control style, camera angles, and character consistency with precision. Cinematic quality with stable, realistic output. Supports complex scenes, professional camera work, multi-shot storytelling, and multiple languages. Perfect for advertising, film, and social media marketing.

Seedance 2.5 is now on Varoriya! Experience 30-second extended storytelling, expanded multimodal references, and total prompt freedom with unrestricted capabilities.

The latest text-to-video model from OpenAI, capable of generating cinematic-quality motion with realistic physics, natural scene continuity, and synchronized sound.It supports prompts to define scenes, voices and dialogues. Use an image for tighter control Add a reference image (photo, digital art, or AI-made) to lock in character look, wardrobe, set details, and overall style. The model treats that image as the first-frame anchor; your text prompt tells it what happens next. Rules Content must be suitable for under-18 audiences (an opt-out may come later). No copyrighted characters or copyrighted music, No real people, including public figures, Input images with human faces are currently rejected. Make sure your prompts, reference images, and transcripts follow these rules to avoid failures.

The latest text-to-video model from OpenAI, capable of generating cinematic-quality motion with realistic physics, natural scene continuity, and synchronized sound.It supports prompts to define scenes, voices and dialogues.

Tweening is a video generation mode using AI that creates a smooth, realistic, and naturally continuous animation between two still images — the starting frame and the final frame — by automatically generating the in-between frames

text-to-speech system that converts written text into natural-sounding speech with customizable voices, emotions, pacing, tone, and languages, making it ideal for video narration, online content, and character voice creation without needing to record audio

A lip-sync model that directly links audio with video, enabling mouth movements to align naturally with the sound.

Video Upscaler reconstructs detail and sharpens your video — take a 480p, 720p, or 1080p clip and upscale it anywhere from 720p up to 4K, with frame rates from 24 to 120 fps. Just upload a clip, select the range you want, choose a resolution and frame rate, and let the AI handle the rest.

A powerful AI tool that transforms text-to-video  & images into high-quality videos (Image-to-Video). Its standout feature is the ability to generate and sync audio simultaneously in one process. Users can provide a single image and a text prompt to direct the action and narrative, and the model will produce a video of up to 1080p resolution, lasting either 5 or 10 seconds, with perfectly matched sound. It also allows for uploading custom audio files to serve as a voiceover or soundtrack.

Faster, smaller, more cost-efficient, and environmentally friendly. High-performance image-to-video generation.

MiniMax H3 is an all-around AI video generation model capable of seamlessly integrating text, images, audio, and video. It allows users to freely mix complex instructions in one go, delivering video outputs up to 15 seconds long at up to 2K resolution. Moreover, its operating cost is many times cheaper than typical models on the market

A fast and cost-efficient video generation model that can create videos in multiple ways — from text, images, audio, or image + audio. (For lip-sync videos, please select 1080p to ensure better mouth and lip quality.)

Seed Audio 1.0. that can handle music, sound effects, and dialogue from references for consistent voices. Text to Background Ambient, Text to Music, Text to Dialogue and SFX, Audio Reference / Voice Cloning

Flux Multi-LoRA is an extension of the Flux and LoRA concept, which involves using multiple LoRAs simultaneously within a single prompt.

As an Image-Prompt model, Flux-Redux uses an original image as its primary input to create new, varied versions. It's designed to maintain the fundamental elements of the source image without needing a text prompt

FLUX is a text-to-image AI model developed by BFL. The model was created as a tool to help users quickly and efficiently generate high-quality images

An AI tool/model for Inpainting or Generative Fill, used to seamlessly complete missing parts of an image or modify specific areas while preserving natural context. Simply paint over any region of the image, then describe in text what you want that area to become, and the system will generate realistic results consistent with the original scene.

Image generation using Flux LoRA is the process of creating images by combining a base model (Flux), which excels at high-quality image generation, with a supplementary model (LoRA) that contains deep, specialized data. This combination helps us achieve images with specific, detailed characteristics.

Flux Canny is edge-guided image generation that uses the reference image’s edge map to control structure and composition, while Flux Depth is depth-aware image generation that uses the reference image’s depth information to edit images while preserving spatial relationships.

Flux Hyper uses an 8-step process, generating 2 images per run. The average generation speed is about 2–3 seconds per image.

Dataset Composer is a tool for creating and editing datasets for LoRA training in image and video generation, supporting existing ZIP datasets or new JPG/PNG images, merging multiple datasets, repairing and upscaling images, cropping, and generating captions automatically with Varo-AI.

Dataset Viewer is a tool for viewing and exploring data contained in a zip file that holds a dataset prepared for model training

Dataset Composer is a tool for creating and editing datasets for LoRA training in image generation, supporting existing ZIP datasets or new JPG/PNG images, merging multiple datasets, repairing and upscaling images, cropping, and generating captions automatically with Varo-AI.

Create a custom voice model by training on your uploaded audio samples for use with the VARO-TTS service. This feature is available exclusively for Corporate accounts

A dataset is a collection of images—sometimes accompanied by caption files—used for training LoRA auxiliary models, typically containing character or style examples at resolutions around 768–1024 pixels to enable effective pattern learning.

Import Flux LoRA models from Hugging Face to your account for image generation in FluxLoRA/FluxMulti-LoRA mode

Varo Braille Gen is a mode that generates Braille images using Artificial Intelligence (AI), designed to open new experiences for people with visual impairments. The output can be displayed on a computer as a Braille dot image, and the file can be printed with a Braille embosser so it can be physically touched as raised dots.

Character Breakdown Sheet (CBS) is a service that generates a structured reference sheet of a character from 1–5 source images, breaking down facial features, hairstyle, outfit, and body proportions into clearly labeled panels from multiple angles — designed to be used as a reference for your next AI image or video generation, improving character consistency and reducing issues like face drift, color mismatch, or outfit inconsistency.

Characters Consistent is the ability to generate the same character repeatedly in different situations or poses.

EdgeFlow is an AI-powered process for expanding images or filling empty spaces around the original image. The AI analyzes the image's details to generate extensions that align with its context, such as adding skies, forests, or buildings in the background. This enhances the image's completeness and visual appeal. Additionally, users can customize the additions by specifying their preferences through prompts.

Models create text descriptions or captions from images using large multimodal transformers trained on image–text pairs to understand visual concepts.

Varo Koneko for Adobe Photoshop AI image-editing plugin for Photoshop — select an area, describe the edit, and the result is placed as a new layer. Powered by the Varoriya API. ⬇ Download (.ccx) Get API Key Features Inpaint by selection — select an area (multiple / irregular / lasso) and the result is placed as a non-destructive Smart Object clipped with a soft, feathered mask. Whole-image edit — with no selection, the result fills the full canvas width and the height follows the generated aspect (matches the source, never stretched). Reference images — attach up to 2 reference images (click to add) alongside the canvas edit. Multiple models — Nanobanana Pro, SeeDream 5 Pro, GPT Image 2, each with quality tiers. Native Photoshop (Spectrum) UI with one-click Generate. Requirements Photoshop 24.0.0 or newer (UXP) — Windows or macOS. Adobe Creative Cloud desktop app (installs the plugin). A Varoriya API key (varo_sk_…). Install Download varo-koneko.ccx from the button above. Double-click the .ccx file — Creative Cloud installs it automatically (Windows & Mac). Open Photoshop → Plugins menu → Varo Koneko to open the panel. In the panel, open the menu (☰) → API Key, paste your API key, and Save. Usage Open an image (optionally select an area to inpaint). (Optional) Add up to 2 reference images — click the Reference box to pick files. Choose a model + quality, then type your edit prompt. Click Generate — the result is placed as a new Smart Object layer. Note: Each generation deducts coins based on the model/quality; failed jobs are auto-refunded. Most image jobs finish in seconds (the panel waits up to ~5 minutes). 📌 Follow the latest version & updates on GitHub Releases.

A digital identity registration system on the Blockchain — protecting your work, identity, and data from copying, forgery, and unauthorized AI usage.

An AI tool or model that automatically removes the background from the main subject in an image, leaving a transparent background.

SpyShot reveals multiple hidden perspectives from a single image you upload, as if you could walk around the scene and observe it from different positions and unique angles that were never visible before.

Varo Boxes  is an AI-powered image editing feature that grants users exact control over the size and placement of elements by drawing bounding boxes directly onto an image. Within these custom boundaries, users can input specific text prompts for targeted modifications or type @ to seamlessly insert uploaded reference products exactly where desired. This tool eliminates the unpredictability of standard AI generation, enabling users to flawlessly execute multiple localized edits—such as adjusting eye color, changing lip texture, and placing a specific product—in a single generation step without confusing the AI or altering the rest of the image.

image generation mode that uses black-and-white reference images—such as QR codes, graphics, or logos—as structural guides to control composition

Transforms your uploaded image into an IMAX-style panoramic scene. Using AI, it analyzes the original image and expands the world in every direction, generating a surrounding environment that feels natural and cohesive—with realistic lighting, shadows, textures, and atmosphere.

image generation mode that analyzes the colors, style, and key elements of your uploaded inspiration image to create a new collection of AI-generated variations, with optional text prompts for further creative direction.

Varo MCP Use Varoriya Services Directly via AI Agents Varo MCP is a connector that enables AI Agents like Claude, VS Code Copilot, Cursor, and ChatGPT to use Varoriya's content generation services directly through chat — you just type what you want, and the AI handles calling the models for you. No need to open a web browser or copy-paste. Services available via Varo MCP: Generate Images — z-image, Grok, Nano Banana Pro, Seedream 5 Pro Generate Videos — Seedance 2.0, WAN, Grok Video (with audio) Generate Audio — Music, ambiance, SFX, voiceover (Seed Audio) Blockchain Registration — Certify your work on the Varoriya Registry (Polygon) Coin Management — Check VaroCoin balance, Top up via PromptPay QR Varo MCP operates via the MCP (Model Context Protocol), an open protocol that allows AI Agents to integrate with external tools — Varo MCP is Varoriya's implementation, instantly giving access to our services for any MCP-compatible AI. Installing Varo MCP Varo MCP is provided via a Remote Server at https://mcp.varoriya.com — no additional installation required. Just add this URL to your client and log in with your Varoriya account via OAuth (no API Key needed). Claude Step-by-Step Installation on Claude Desktop Claude Desktop is a popular AI Assistant — once Varo MCP is installed, you can generate images, videos, and audio directly through your Claude chat. Open Claude Desktop Go to Settings (⚙️ bottom left or Ctrl+,) Select the Connectors tab Click Add custom connector Enter URL: https://mcp.varoriya.com Click Connect → An OAuth window will open for you to log in with your Varoriya account Once connected successfully, you will see varo in the Connectors list with a Connected ✓ status — It is recommended select Always allow so Claude can call the tool instantly without asking every time. After installing, try typing one of the examples below → Chat Prompt Examples Tip: If your Claude Desktop does not have a Connectors tab yet, update it to the latest version, or view the stdio bridge installation guide at dev.varoriya.com/mcp Chat Prompt Examples Once Varo MCP is installed, you can instruct the AI immediately. The AI will automatically call the relevant tools — you don't need to specify the tool name yourself. Generate Images "Generate a 16:9 picture of a sea of mist in the morning using z-image" "Draw a 3D cartoon style robot watering plants using Grok" "Create an image of cosmetic products on a marble background in 1K size using Seedream 5 Pro" "Edit this picture to make the whole image blue using Nano Banana Pro" (attach original image) "Modify this photo by changing the shirt to a formal suit with a plain background, keep the original face strictly unchanged" (attach original image) "Restore this old torn photo of my grandmother, fix the tears, keep the original face strictly unchanged using gptimage" (attach original image) Generate Videos "Make a 5-second video of a waterfall in a tropical rainforest with sunlight shining through leaves using Seedance 2.0" "Create an 8-second video from this picture, make the character turn to smile and wave using WAN" "Make a 4-second cinematic video of a cat walking on a fence using Grok Video" "Use this photo to make a video with Grok Video, making them say 'Hello' in Thai" (attach original image) Generate Audio "Create a beautiful 60-second Lo-fi piano instrumental for studying" "Generate 30 seconds of tropical rainforest nature sounds with birdsong and flowing water" "Create a Thai female voiceover with a warm tone saying 'Welcome to Varoriya'" Check Balance / Top Up "Please check my remaining coin balance" "Top up 300 Baht" "I want to top up coins, what packages are available?" View Models and Pricing "What models are available to use right now?" "Check the price for 2K quality z-image" "How many coins does generating an image with z-image cost?" "Which video generation model is the cheapest?" Blockchain Registration "Check the registration price for this image on the Blockchain" "Register this image on the Varoriya Registry" "This picture is amazing, register it on the blockchain" Coin Management Pre-deduct on submit — If a job fails, coins are automatically refunded Check remaining balance: varo_balance Top up coins: varo_topup — Get a PromptPay QR code to scan and pay Request an API Key for Direct API usage: varoriya.com/account/key View Generation History Go to varoriya.com/api-usage — displays your entire generation history via MCP and API, categorized by app (MCP: VS Code, MCP: Cursor, MCP: Claude, API direct) Frequently Asked Questions (FAQ) Do I need an API Key? No — Varo MCP uses OAuth to log in directly with your Varoriya account. No need to copy-paste any keys. If you want to use an API Key (e.g., for CI/CD or automated systems), you can request one at varoriya.com/account/key and add it as a header: Authorization: Bearer varo_gk_... How long are generated files kept? Output files (images/videos/audio) are kept for 24 hours on the VARORIYA CDN — which is sufficient time for downloading. If you want permanent storage, register them on the Varoriya Registry. Which Clients are supported? Client Remote HTTP stdio bridge Deeplink Notes Claude Desktop ✅ ✅ ❌ Connector is recommended (Settings → Connectors) VS Code ✅ ✅ ✅ One-click install Cursor ✅ ✅ ✅ One-click install ChatGPT ✅ ❌ ❌ Business / Enterprise / Edu plans only (beta) Claude Code (CLI) ✅ ❌ ❌ claude mcp add varo --url ... Which APIs does MCP use under the hood? Varo MCP calls the Varo Generate API (/v1/generate/image, /v1/generate/video, /v1/generate/audio) — view the full documentation at the API Reference Related Links Link Description dev.varoriya.com Developer Hub — Tools for developers dev.varoriya.com/mcp MCP Setup Page (One-click install) dev.varoriya.com/reference Generate API Reference (OpenAPI/Scalar) dev.varoriya.com/playground Playground — Try the Generate API in your browser api.varoriya.com REST API Endpoint mcp.varoriya.com MCP Server Endpoint varoriya.com/account/key Request an API Key varoriya.com/api-usage View generation history via API/MCP github.com/varoriya/varo-mcp Source Code (GitHub)

Create 360-degree images with AI from either existing images or text prompts, allowing viewers to look around in every direction as if they were standing at that location. This is ideal for wide-angle scenes, locations, and landscapes. If a reference image includes a person, it is recommended to use a full-body image so the system can generate the surrounding environment more naturally. The system supports multiple reference images, enabling the AI to use them as context for generating a complete 360° environment. The final result can also be posted and viewed directly as a 360° image on Facebook

VaroChrome 2.0: The All-in-One AI Extension to Elevate Generative AI Workflows Overview Working with AI tools today often involves complex steps and switching between multiple platforms—from finding references and adjusting prompts to generating images and uploading them for video creation. VaroChrome 2.0 simplifies this process, instantly transforming text or images into still images, videos, and voiceovers. It seamlessly connects prompt creation with precise post-processing all in one place, without leaving the current web page. Key Features Seamless Integration: Connects directly with ChatGPT, Gemini, and Pinterest. The VaroChrome icon automatically appears under images. Simply click to configure, edit, or generate videos instantly. Universal Access: For other websites, simply "right-click" on any image and select the VaroChrome menu. The extension panel will open, ready for configuration and generation. Top-Tier AI Models: Image Models: Seedream 5 Pro, Nano Banana Pro, GPT-Image 2.0, Grok Image, Z-Image Video Models: Seedance 2.0, WAN 2.2, H3, Flux 3, including Grok Video with Thai speech support Audio Models: Seed Audio in 19 languages  for speech, music, and sound effects generation How to Use Install: Download and install VaroChrome on the browser. Log in: Connect a Varoriya account (one-click) or provide a Generation API Key. Select Reference: Open any desired website. Activate: Right-click on an image → select "VaroChrome Panel" (or click the VaroChrome icon under images on supported sites). Create: Select a model, adjust settings as needed, and click Generate. Download: Save the results for immediate use. Privacy Policy Highest priority on user data security: Local Processing: All data is handled locally on the device until a media generation command is sent. Auto-Deletion: Media files generated by the system are automatically deleted from the server within 24 hours. Download VaroChrome 2.0: Chrome Web Store