# dGEN Visual Studio

> Generate images and video with AI. Three ways to work: a straightforward generator,
> an assistant that sets it up for you, or a canvas where you connect prompts, models,
> reference images and your own LoRA into a workflow you save and run as often as you
> like.

## What it does

- Text to image, image to image, and your own LoRA to image
- Text to video, image to video, video to video, and your own LoRA to video
- First frame to last frame, and reference images that keep one character across shots
- dGEN Influencer: build a character from 20 groups of picture tiles, get a portrait, a full body and a character sheet, save it, make its first vertical Reel in one click, and use the same face in new videos, swap it into a clip you have, or let it talk
- Camera, lens, light and look presets that are written into your prompt word for word
- Every generation priced in credits, with the cost shown before you start
- One-off generations, or an assistant that writes the prompt. You pick the model, or leave it to the assistant
- Bring your own style: upload a LoRA or paste a Hugging Face or Civitai link
- Build a workflow once, then reuse, duplicate or share it as a template
- Full refund if a run fails before the model has made anything, or through an outage

## Where to read more

- dGEN Studio (https://dgenvisual.com/studio): the whole studio in one screen, every model for image and video.
- dGEN AI Image Generator (https://dgenvisual.com/ai-image-generator): Text to image, image to image, and your own LoRA.
- dGEN AI Video Generator (https://dgenvisual.com/ai-video-generator): Text to video, image to video and video to video, with sound where the model supports it.
- dGEN AI Assistant (https://dgenvisual.com/ai-assistant): Describe what you want and it writes the prompts. Name a model and it uses that one, or leave the choice to it.
- dGEN Influencer (https://dgenvisual.com/ai-influencer-generator): Build a character from tiles, keep the same face in every image and video.
- dGEN Flow (https://dgenvisual.com/flow): A node canvas that turns a one-off into a repeatable pipeline.
- dGEN Templates (https://dgenvisual.com/templates): ready-made flows that already do the work.
- Pricing (https://dgenvisual.com/pricing): credit packs, no subscription, credits never expire.
- Showcase (https://dgenvisual.com/showcase): real results from the studio, with the model and the prompt.

## Who it is for

- E-commerce (https://dgenvisual.com/for/ecommerce): Product photos and video for your webshop, from one product photo.
- Agencies (https://dgenvisual.com/for/agencies): Every model for every client in one studio, with pipelines you can run again.
- Content creators (https://dgenvisual.com/for/content-creators): Visuals and clips in your own style, every day, with one character across all of them.
- Filmmakers (https://dgenvisual.com/for/filmmakers): Previs, storyboards and shots that share one look, with the camera moves you choose.
- Musicians (https://dgenvisual.com/for/musicians): A music video for your own track: your artist performs every line, close up, with the mouth on the vocal.

## Tools, one screen per job

- Expand image (https://dgenvisual.com/tools/expand-image): Widen a picture beyond its edges. Pick the shape you need, no prompt required. You bring: One image.
- Transition between two frames (https://dgenvisual.com/tools/transition): Give a first frame and a last frame, and get the clip that runs between them. You bring: Two images: where it starts and where it ends.
- Keep the same character (https://dgenvisual.com/tools/same-character): Reference images that carry one person, product or place across a whole sequence. You bring: One to five reference images of the thing that has to stay the same.
- Remove an image background (https://dgenvisual.com/tools/remove-background): Cut the subject out of a photo and leave the background transparent. You bring: One image.
- Remove a video background (https://dgenvisual.com/tools/remove-video-background): Cut the subject out of a clip and leave the background transparent. You bring: One video.
- Face swap in a clip (https://dgenvisual.com/tools/face-swap): Put one face into an existing video. Only with permission from the person in the picture. You bring: A clip, and a clear photo of the face that goes in.
- Make your character move like you (https://dgenvisual.com/tools/motion-control): Film a move once, and your character does it: the same steps, the same timing. You bring: A clip of the move, with the person fully in frame, and a full-body image of your character.
- Recast a video (https://dgenvisual.com/tools/recast): Swap the person on screen for someone in a photo. The motion, the cuts and the sound stay. You bring: A video up to 30 seconds, and one clear photo per new person, up to four.
- Extend a video (https://dgenvisual.com/tools/extend-video): Make a clip longer. Describe what happens next and the video carries on in the same style. You bring: A video, and a sentence about what happens next.
- Make a photo talk (https://dgenvisual.com/tools/talking-photo): Give a face in a photo your audio. The lips follow the words. You bring: A photo with a clear face, and an MP3 or WAV.
## Models, what each one is for

- Nano Banana Pro by Google DeepMind (https://dgenvisual.com/models/nano-banana-pro): Studio-grade images with text that actually reads, up to 4K. Best for: Posters, packaging and ads where the text has to be right.
- Seedance by ByteDance (https://dgenvisual.com/models/seedance): Multi-shot video from text or an image, in 1080p. Best for: Multi-shot sequences with big movement.
- FLUX.2 by Black Forest Labs (https://dgenvisual.com/models/flux-2): Production-grade images up to 4MP, with up to ten reference images. Best for: Product shots that match up to ten references.
- Nano Banana 2.1 by Google DeepMind (https://dgenvisual.com/models/nano-banana-2): Fast image generation and editing, now up to 4K. Best for: Fast edits and quick rounds of changes, up to 4K.
- Nano Banana by Google DeepMind (https://dgenvisual.com/models/nano-banana): The original Gemini image model. Cheap, quick, good enough to explore with. Best for: Cheap first drafts while you find the idea.
- Kling 4.0 by Kling AI (https://dgenvisual.com/models/kling-4): The newest Kling: video up to 30 seconds, several shots in one run, with sound. Best for: Long clips of up to 30 seconds, told in several shots.
- Kling 3.0 by Kling AI (https://dgenvisual.com/models/kling): Cinematic video up to 15 seconds, with native audio and named elements. Best for: Cinematic clips up to 15 seconds with a recurring character.
- Seedream 5 by ByteDance (https://dgenvisual.com/models/seedream): Dense layouts and real typography, in fourteen languages. Best for: Layouts, infographics and type in many languages.
- GPT Image 2.5 by OpenAI (https://dgenvisual.com/models/gpt-image-2): Near-perfect text in images, and it follows a long brief. Best for: Long, precise briefs and text you do not have to fix.
- MiniMax H3 Max by fal Research (https://dgenvisual.com/models/h3-max): Fast video, and a toolkit around it: recast, extend, insert, lip sync. Best for: Recasting, extending and lip-syncing footage you already have.
- Sync Lipsync 2 by Sync Labs (https://dgenvisual.com/models/sync-lipsync): Your video, your audio, the mouth exactly on it. Best for: Dubbing: a new voice on footage you already have, the mouth on the audio frame by frame.
- InfiniteTalk by MeiGen-AI (https://dgenvisual.com/models/infinitalk): Lip sync on a video you have, with the head and face moving on the audio. Best for: Short talking clips where the head should move with the words.
- OmniHuman 2 by ByteDance (https://dgenvisual.com/models/omnihuman): Your photo performs your track, in one take. Best for: Music videos and talking takes where the whole person moves on the audio.
- Mirelo SFX 2 by Mirelo (https://dgenvisual.com/models/mirelo-sfx): Sound effects that land on what happens in your video. Best for: Impacts, steps, weather and machines, timed to your picture.
- MiniMax H3 by MiniMax (https://dgenvisual.com/models/minimax-h3): A frontier video model that runs from 480p drafts up to 4K. Best for: Video from cheap 480p drafts to 4K finals.
- Ideogram by Ideogram (https://dgenvisual.com/models/ideogram): Posters, logos and readable type, and precise edits that keep the rest intact. Best for: Posters, logos and edits that leave the rest alone.
- PixVerse V6 by PixVerse (https://dgenvisual.com/models/pixverse): Four resolutions from cheap drafts to 1080p, with optional sound. Best for: Cheap video drafts, with sound when you need it.
- LTX 2.5 by Lightricks (https://dgenvisual.com/models/ltx-2-5): Video and audio generated together in one pass, from a shot brief. Best for: Clips with picture and sound from one shot brief.
- Wan 2.7 by Alibaba (https://dgenvisual.com/models/wan): Smooth motion and scenes that stay coherent, from a prompt alone. Best for: Coherent motion from a prompt, no start image needed.
- Wan 2.2 Animate by Alibaba (https://dgenvisual.com/models/wan-animate): Put your character in a clip you already have, or give it the moves from that clip. Best for: Swapping a character into your own clip, at a low price.
- Topaz by Topaz Labs (https://dgenvisual.com/models/topaz): Professional upscaling, up to 4x, faithful to the original. Best for: Upscaling finished work for print and big screens.
- Z-Image Turbo by Tongyi-MAI (https://dgenvisual.com/models/z-image-turbo): Three LoRAs at once, at roughly a hundred generations per dollar. Best for: High volume with your own LoRAs, at a low price.
- Qwen Image by Alibaba (https://dgenvisual.com/models/qwen-image): Better text, finer textures, and your own LoRA on top. Best for: Your own LoRA, with clean text and fine detail.
- Krea 3 by Krea (https://dgenvisual.com/models/krea): Fast, precise image edits that leave alone what you did not ask to change. Best for: Product scenes and edits that keep the label intact.
- Grok Imagine by xAI (https://dgenvisual.com/models/grok-imagine): Image and video from xAI, stylised rather than photoreal. Best for: Stylised images and clips with a look of their own.
- FLUX.1 dev by Black Forest Labs (https://dgenvisual.com/models/flux-1-dev): The model most LoRAs on the internet were trained on. Best for: Running the LoRAs you find online.
- FLUX 3 by Black Forest Labs (https://dgenvisual.com/models/flux-3): Black Forest Labs' newest family: images up to 4K, and video with sound. Best for: 4K stills and long clips from one family.
- Vidu by Vidu (https://dgenvisual.com/models/vidu): Image to video from cheap 360p drafts to 4K with sound in Q4. Best for: Testing image-to-video ideas cheaply.
- Wan 3.0 by Alibaba (https://dgenvisual.com/models/wan-3): Start and end frames, its own audio, and motion that stays coherent. Best for: Shots that run from your start frame to your end frame.
- Gemini Omni Flash 1.1 by Google DeepMind (https://dgenvisual.com/models/gemini-omni-flash): Video with sound made in the same pass, from a frame, a prompt or a reference. Best for: A character that keeps its look and voice across clips.
- Bria by Bria (https://dgenvisual.com/models/bria): Two utilities: make an image bigger, or take the background out of a clip. Best for: Bigger images, and backgrounds out of video.
- SeedVR2 by ByteDance (https://dgenvisual.com/models/seedvr): Makes a clip sharper without the frames drifting apart. Best for: Sharper clips without flicker.
- BiRefNet by BiRefNet (https://dgenvisual.com/models/birefnet): Takes the background out of an image, edge and all. Best for: Clean cut-outs for product photos.
- Fast upscale by Chrono Edit (https://dgenvisual.com/models/chrono-edit): A cheap way to clean up an image, when Topaz is more than you need. Best for: Quick, cheap image clean-ups.
- Kandinsky 6 by Kandinsky Lab (https://dgenvisual.com/models/kandinsky-6): Cinematic video from a long prompt, and an upscaler that keeps frames steady. Best for: Cinematic clips from long prompts, and steady video upscaling.
- Recraft 4.1 Flash by Recraft (https://dgenvisual.com/models/recraft): Images for under a cent, with your own colour palette. Best for: Cheap drafts, icons and flat illustration in your colours.
- MOSS Image by OpenMOSS (https://dgenvisual.com/models/moss-image): Images with readable English and Chinese text, laid out where you ask. Best for: Posters, signs and layouts with text in English or Chinese.
- Bria FIBO 2.0 by Bria (https://dgenvisual.com/models/bria-fibo-2): Images and edits from a model trained only on licensed data, with clear commercial rights. Best for: Commercial images where the rights have to be clean.
- Bria FIBO Edit by Bria (https://dgenvisual.com/models/bria-fibo): Put your product in someone's hand, or your clothes on them. Best for: Product and clothing photos for a webshop.

## What is new

- 2026-10-24: Fourteen new AI influencer examples. A news anchor with bulging eyes, a crocodile barista, a skateboarding grandma, a Bigfoot vlogger, an octopus barber and nine more, each with portrait, full body and character sheet. Press Recreate on one and the builder starts from the same choices. (https://dgenvisual.com/ai-influencer-generator)
- 2026-10-23: Template: your influencer talks. Put your character in a cinematic 8-second vertical clip where it speaks one line in its own voice, with lip sync and the sound of a real-looking place. The template behind our AI influencer commercial. (https://dgenvisual.com/templates/talking-ai-influencer)
- 2026-10-23: Every model now costs about half. A run now costs about half the credits it did. An eight-second clip with sound on Gemini Omni Flash 1.1 at 1080p is 2,400 credits, and a character with portrait, full body and sheet costs 497. The price is still on the button before you press it. (https://dgenvisual.com/pricing)
- 2026-10-22: Compare any two models, and the best model per job. Fifteen side-by-side comparisons, such as Kling vs Seedance and Nano Banana Pro vs FLUX, with real prices, and guides that name the best model for video, images, text, product photos, lip sync and more. (https://dgenvisual.com/compare)
- 2026-10-22: A page for every template. Each of the twenty templates now has its own page: what it makes, the models in it and how to use it. (https://dgenvisual.com/templates)
- 2026-10-22: Ideogram: replace a background, remix, transparent images. Three new Ideogram v3 options: put a subject on a new background, remix an image in a new direction, and generate images with a transparent background. (https://dgenvisual.com/models/ideogram)
- Full changelog: https://dgenvisual.com/changelog

## Comparisons

- Kling vs Seedance: which AI video model to use (https://dgenvisual.com/compare/kling-vs-seedance): Pick Kling 3.0 for one longer cinematic shot with sound and a named character it keeps. Pick Seedance for several shots that belong together in one clip and for big movement.
- Kling vs Hailuo (MiniMax H3): AI video compared (https://dgenvisual.com/compare/kling-vs-hailuo): Pick Kling 3.0 for long cinematic shots with sound and a character it keeps. Pick MiniMax H3 (Hailuo) when you want to draft cheaply at 480p and finish the same shot in 4K.
- Seedance vs Hailuo (MiniMax H3): which is better (https://dgenvisual.com/compare/seedance-vs-hailuo): Pick Seedance for a sequence of shots with big movement in one clip. Pick MiniMax H3 (Hailuo) to control both ends of a shot and to go from cheap drafts to 4K.
- Kling vs Wan 2.7: AI video models compared (https://dgenvisual.com/compare/kling-vs-wan): Pick Kling 3.0 for longer cinematic shots with sound and a character it keeps. Pick Wan 2.7 for coherent motion straight from a prompt, without a start image.
- Kling vs Gemini Omni Flash: AI video with sound (https://dgenvisual.com/compare/kling-vs-gemini-omni-flash): Both make sound with the picture. Pick Kling 3.0 for longer cinematic shots. Pick Gemini Omni Flash for a character that keeps its look and its voice across clips, and for cheap runs.
- Gemini Omni Flash vs LTX 2.5: video and sound together (https://dgenvisual.com/compare/gemini-omni-flash-vs-ltx): Pick Gemini Omni Flash for cheap exploring and a character that keeps its voice. Pick LTX 2.5 for the final pass of a shot you already found, written as a shot brief.
- Nano Banana Pro vs FLUX.2: AI image models compared (https://dgenvisual.com/compare/nano-banana-pro-vs-flux): Pick Nano Banana Pro for text that has to be right and edits that behave like a camera. Pick FLUX.2 to match up to ten reference images and to run your own LoRA.
- Nano Banana Pro vs GPT Image: text and edits (https://dgenvisual.com/compare/nano-banana-pro-vs-gpt-image): Both set text well. Pick GPT Image 2.5 for a long, precise brief it reads to the end. Pick Nano Banana Pro for a cast that stays the same and edits that behave like a camera.
- Nano Banana Pro vs Nano Banana 2: which to use (https://dgenvisual.com/compare/nano-banana-pro-vs-nano-banana-2): Pick Nano Banana 2.1 for fast edits and quick rounds of changes. Pick Nano Banana Pro for text on a poster and for edits that have to hold across a series.
- Seedream 5 vs Nano Banana Pro: layouts and text (https://dgenvisual.com/compare/seedream-vs-nano-banana-pro): Pick Seedream 5 for dense layouts and infographics, and for type in many languages. Pick Nano Banana Pro for editing an image you have and a cast that stays the same.
- Ideogram vs GPT Image: AI images with readable text (https://dgenvisual.com/compare/ideogram-vs-gpt-image): Pick Ideogram for posters, logos and edits that leave the rest alone, at one price whatever the size. Pick GPT Image 2.5 for a long, precise brief with text you do not want to fix.
- Ideogram vs Recraft: AI design images compared (https://dgenvisual.com/compare/ideogram-vs-recraft): Pick Ideogram when the text has to read: posters, logos, packaging. Pick Recraft 4.1 Flash for cheap drafts, icons and flat illustration in your own colour palette.
- Wan 2.2 Animate vs Kling motion control (https://dgenvisual.com/compare/wan-animate-vs-kling-motion-control): Pick Wan 2.2 Animate to swap your character into a clip you filmed, cheaply, with your clip staying your clip. Pick Kling motion control to give the character from your photo the moves from your video.
- Sync Lipsync vs InfiniteTalk: AI lip sync compared (https://dgenvisual.com/compare/sync-lipsync-vs-infinitalk): Pick Sync Lipsync 2 for dubbing: the mouth on the audio frame by frame, the rest of your shot untouched. Pick InfiniteTalk for short talking clips where the head should move with the words.
- Topaz vs a fast AI upscaler: which to use (https://dgenvisual.com/compare/topaz-vs-fast-upscale): Pick Topaz for finished work that goes to print or a big screen: bigger, not different. Pick the fast upscaler for a quick, cheap clean-up when print size is not the goal.

## Which model for which job

- Video from a prompt or an image: Kling 4.0 (https://dgenvisual.com/models/kling-4), Kling 3.0 (https://dgenvisual.com/models/kling), Seedance (https://dgenvisual.com/models/seedance), MiniMax H3 (https://dgenvisual.com/models/minimax-h3), Wan 2.7 (https://dgenvisual.com/models/wan), Wan 3.0 (https://dgenvisual.com/models/wan-3), Gemini Omni Flash 1.1 (https://dgenvisual.com/models/gemini-omni-flash), LTX 2.5 (https://dgenvisual.com/models/ltx-2-5), PixVerse V6 (https://dgenvisual.com/models/pixverse), Vidu (https://dgenvisual.com/models/vidu), Kandinsky 6 (https://dgenvisual.com/models/kandinsky-6), FLUX 3 (https://dgenvisual.com/models/flux-3), Grok Imagine (https://dgenvisual.com/models/grok-imagine), MiniMax H3 Max (https://dgenvisual.com/models/h3-max)
- Images from a prompt: Nano Banana Pro (https://dgenvisual.com/models/nano-banana-pro), Nano Banana 2.1 (https://dgenvisual.com/models/nano-banana-2), Nano Banana (https://dgenvisual.com/models/nano-banana), FLUX.2 (https://dgenvisual.com/models/flux-2), Seedream 5 (https://dgenvisual.com/models/seedream), GPT Image 2.5 (https://dgenvisual.com/models/gpt-image-2), Ideogram (https://dgenvisual.com/models/ideogram), Qwen Image (https://dgenvisual.com/models/qwen-image), FLUX 3 (https://dgenvisual.com/models/flux-3), Recraft 4.1 Flash (https://dgenvisual.com/models/recraft), Z-Image Turbo (https://dgenvisual.com/models/z-image-turbo), FLUX.1 dev (https://dgenvisual.com/models/flux-1-dev), Grok Imagine (https://dgenvisual.com/models/grok-imagine), Krea 3 (https://dgenvisual.com/models/krea), Bria FIBO 2.0 (https://dgenvisual.com/models/bria-fibo-2), MOSS Image (https://dgenvisual.com/models/moss-image)
- Readable text in an image (posters, packaging, logos): Ideogram (https://dgenvisual.com/models/ideogram), GPT Image 2.5 (https://dgenvisual.com/models/gpt-image-2), Nano Banana Pro (https://dgenvisual.com/models/nano-banana-pro), Seedream 5 (https://dgenvisual.com/models/seedream), Qwen Image (https://dgenvisual.com/models/qwen-image)
- Editing an image you already have: Krea 3 (https://dgenvisual.com/models/krea), Nano Banana Pro (https://dgenvisual.com/models/nano-banana-pro), Nano Banana 2.1 (https://dgenvisual.com/models/nano-banana-2), Bria FIBO 2.0 (https://dgenvisual.com/models/bria-fibo-2), Bria FIBO Edit (https://dgenvisual.com/models/bria-fibo), FLUX.2 (https://dgenvisual.com/models/flux-2), Ideogram (https://dgenvisual.com/models/ideogram), GPT Image 2.5 (https://dgenvisual.com/models/gpt-image-2)
- Your own trained style (LoRA): FLUX.1 dev (https://dgenvisual.com/models/flux-1-dev), Z-Image Turbo (https://dgenvisual.com/models/z-image-turbo), Qwen Image (https://dgenvisual.com/models/qwen-image), FLUX.2 (https://dgenvisual.com/models/flux-2)
- One character across clips, or put into a video you have: Wan 2.2 Animate (https://dgenvisual.com/models/wan-animate), MiniMax H3 Max (https://dgenvisual.com/models/h3-max), Kling 3.0 (https://dgenvisual.com/models/kling), InfiniteTalk (https://dgenvisual.com/models/infinitalk), OmniHuman 2 (https://dgenvisual.com/models/omnihuman)
- Lip sync on a video you already have: Sync Lipsync 2 (https://dgenvisual.com/models/sync-lipsync), InfiniteTalk (https://dgenvisual.com/models/infinitalk), MiniMax H3 Max (https://dgenvisual.com/models/h3-max), Wan 2.2 Animate (https://dgenvisual.com/models/wan-animate), OmniHuman 2 (https://dgenvisual.com/models/omnihuman)
- Upscaling images and video: Topaz (https://dgenvisual.com/models/topaz), SeedVR2 (https://dgenvisual.com/models/seedvr), Fast upscale (https://dgenvisual.com/models/chrono-edit), Bria (https://dgenvisual.com/models/bria), Kandinsky 6 (https://dgenvisual.com/models/kandinsky-6)
- Removing a background from an image or video: BiRefNet (https://dgenvisual.com/models/birefnet), Bria (https://dgenvisual.com/models/bria)


## Guides

- Make an image, then change it (https://dgenvisual.com/guides/make-it-then-change-it): The second attempt is usually an edit, not a new generation. Knowing which is which saves most of the cost.
- Text that actually reads (https://dgenvisual.com/guides/text-in-an-image): Most image models approximate letter shapes. A few set real type, and that decides which one you pick.
- Backgrounds, out and back in (https://dgenvisual.com/guides/removing-a-background): Cutting a subject out is its own kind of model, and the edge is the whole job.
- Making something bigger (https://dgenvisual.com/guides/making-something-bigger): Upscalers do not all do the same thing, and one of them invents detail that was not there.
- Build it once, run it often (https://dgenvisual.com/guides/build-it-once): Retyping the same series every week is the part that costs the time. A pipeline is drawn once.
- Start halfway, not from an empty canvas (https://dgenvisual.com/guides/starting-from-a-template): A template is a working pipeline with the models, the connections and the settings already chosen.
- Your own trained style (https://dgenvisual.com/guides/your-own-style): A reference holds a subject for a few shots. A trained style holds it for as long as you keep using it.
- From an idea to a clip (https://dgenvisual.com/guides/from-an-idea-to-a-clip): The four ways a video model can be started, and how to tell which one your idea needs.
- Start and end frames (https://dgenvisual.com/guides/start-and-end-frames): Hand a model both ends of a movement and it fills in the middle. Here is when that beats a prompt.
- One subject across shots (https://dgenvisual.com/guides/references-in-video): A prompt cannot hold a face steady. A reference can, and several models take images, clips and audio together.
- Sound made with the picture (https://dgenvisual.com/guides/sound-made-with-the-picture): Some models generate audio in the same pass as the image. That is a different thing from adding a track afterwards.
- Clips that follow on (https://dgenvisual.com/guides/clips-that-follow-on): No model here makes a two-minute film. Several short clips that share their frames do.
- How to write a prompt that works (https://dgenvisual.com/guides/writing-prompts): The same idea gets a different result on every model. This is why, and what to write instead.
- Camera, lens and light, in words a model understands (https://dgenvisual.com/guides/camera-and-lens): The vocabulary a cinematographer uses, and what each term does to a generated shot.
- Which model for which job (https://dgenvisual.com/guides/choosing-a-model): 199 models is too many to choose from. Here is how the choice actually narrows.
- Sizes and aspect ratios (https://dgenvisual.com/guides/aspect-ratios): Why the same shape has two different names, and how to be sure you get the one you asked for.
- Credits, and what a generation actually costs (https://dgenvisual.com/guides/credits-and-cost): Where the number on the button comes from, and what happens when a run fails.
- A character that keeps coming back (https://dgenvisual.com/guides/a-character-that-keeps-coming-back): How to lock one person once and put them in scene after scene, without a family resemblance every time.
- Build an AI influencer people recognise (https://dgenvisual.com/guides/build-an-ai-influencer): From a first look to your first Reel: build a character once and keep the same face in every post.
- Your AI writer can make its own images (https://dgenvisual.com/guides/blog-images-from-your-agent): Posts go out without a picture because the thing writing them cannot see. Here is how to fix that.
- A different person in the same video (https://dgenvisual.com/guides/recast-a-video): Film once, cast as often as you like. The motion, the cuts and the sound stay exactly as they were.
- A music video for your own track (https://dgenvisual.com/guides/making-a-music-video): Your artist performs your track, close up and in motion, with the mouth on every word.
- Video you can look around in (https://dgenvisual.com/guides/video-in-360): One clip that holds the whole sphere around the viewer, for a headset, a 360 player or a scrollable embed.
- Working on footage you already have (https://dgenvisual.com/guides/working-on-footage-you-have): Make a clip longer, put a scene in the middle, sharpen it, or make a face speak, without starting over.
- The best AI model for every job (https://dgenvisual.com/guides/best-ai-model-for-every-job): Video, images, text in images, product photos, lip sync, upscaling: which AI model does each job best, and why.
- The best AI video generator for each kind of shot (https://dgenvisual.com/guides/best-ai-video-generator): Kling, Seedance, MiniMax H3, Gemini Omni Flash, Wan and LTX: which AI video model wins which shot, and why.
- The best AI image generator for each kind of picture (https://dgenvisual.com/guides/best-ai-image-generator): Nano Banana Pro, FLUX.2, GPT Image 2.5, Seedream 5, Ideogram and Recraft: which image model to use for what.
- The best image to video AI for your photo (https://dgenvisual.com/guides/best-image-to-video-ai): Turning a photo into a moving shot: which AI model holds your image best, steers the ending, or tests ideas cheaply.
- The best AI image generator for readable text (https://dgenvisual.com/guides/best-ai-for-text-in-images): Posters, packaging, social posts and logos: the AI image models that set real type, and which one to use when.
- The best AI for product photos of your real product (https://dgenvisual.com/guides/best-ai-for-product-photos): Lifestyle shots, a product in someone's hand, clothes on a model: the AI models that keep your product your product.
- The best AI lip sync for your video or photo (https://dgenvisual.com/guides/best-ai-lip-sync): Dubbing footage, a head that moves with the words, or a photo that speaks: which AI lip sync to use for which job.
- The best AI upscaler for images and video (https://dgenvisual.com/guides/best-ai-upscaler): Making an image bigger or a clip sharper: which AI upscaler stays true, which is cheap, and which keeps video steady.
- The best AI for a consistent character (https://dgenvisual.com/guides/best-ai-for-consistent-characters): The same face in every image and clip: how to build a character once and which AI models carry it best.
- The best AI photo editor for changing one thing (https://dgenvisual.com/guides/best-ai-photo-editor): Changing a colour, swapping a background, removing an object: the AI editors that change what you ask and keep the rest.
- The best AI music video generator for your track (https://dgenvisual.com/guides/best-ai-music-video-generator): Your artist performs your track with the mouth on the vocal: the AI models for each step of a music video.
- Design an AI character from scratch (https://dgenvisual.com/guides/ai-character-generator): Head shape, face, hair, outfit and a signature gesture: how to design a character that is yours and stays recognisable.
- Your first Instagram Reel with an AI influencer (https://dgenvisual.com/guides/first-reel-with-an-ai-influencer): A ten-second vertical clip with a hook, a scene and your character's signature gesture, with sound, from the character you built.
- Face swap a video with AI, the right way (https://dgenvisual.com/guides/face-swap-a-video): Put one face into an existing clip: what to bring, when to use recast instead, and why permission comes first.
- AI motion control: your character copies your moves (https://dgenvisual.com/guides/ai-motion-control): Film a move once and your character does it, with the same steps and timing. Which of the three ways to use, and how to film it.
- Wan 2.2 Animate: put your character in your own video (https://dgenvisual.com/guides/wan-2-2-animate): Swap your character into a clip you filmed, or give your character the moves from it. No prompt, a low price per try.
- Kling image to video: how to get the best clip (https://dgenvisual.com/guides/kling-image-to-video): How to prompt Kling 3.0 from a photo: motion first, a start and an end, sound, a character it keeps, and which tier to pick.
- Nano Banana prompts that work, with examples (https://dgenvisual.com/guides/nano-banana-prompts): How to write a prompt for Nano Banana Pro and 2.1: subject, setting, light and finish, text in quotes, and edits that change one thing.
- Make a photo talk or sing with AI (https://dgenvisual.com/guides/talking-photo-ai): One photo and your audio, and the face speaks or sings it. What works, what to expect, and when to use lip sync on video instead.
- AI UGC video from one product photo (https://dgenvisual.com/guides/ai-ugc-video): Turn a product photo into a clip that looks filmed by someone who owns it, through three video models at once, and pick the take that sells.
- Make images and video from Claude with MCP (https://dgenvisual.com/guides/claude-mcp-images-and-video): Connect Claude, Claude Code or your own agent to the studio over MCP, and let it make images and video with the price shown first.

## Questions people ask

- Frequently asked questions: https://dgenvisual.com/faq
- About the company: https://dgenvisual.com/about
- Contact, teams and invoicing: https://dgenvisual.com/contact
- Terms, licence, acceptable use, privacy and copyright: https://dgenvisual.com/legal
- We claim no ownership of your results and do not train on your work.
- Every model in the studio is licensed by its provider for commercial use.
- Copyright in purely AI-generated material is not a given under European law.

## Blocks on the canvas

- Prompt
- Prompt assistant
- File
- Image generator
- Video generator
- Edit
- LoRA
- Character

## Starting points

- Character Sheet (https://dgenvisual.com/templates/character-sheet): One photo in, a four-panel reference sheet of that same person out.
- AI influencer sheet (https://dgenvisual.com/templates/continuity-sheet): One photo in, a full continuity sheet out: profile, turnaround, expressions, poses and details, with labels.
- 8 Camera Angles (https://dgenvisual.com/templates/camera-angles): One image in, the same subject from eight camera positions out.
- Product visual (https://dgenvisual.com/templates/product-visual): Your product photo in a styled scene, with its shape intact.
- Before and after (https://dgenvisual.com/templates/before-and-after): Show a room before and after a restyling.
- Product to UGC (https://dgenvisual.com/templates/product-to-ugc): Turn a product photo into authentic UGC content, in three models at once.
- Multi-Model Video (https://dgenvisual.com/templates/compare-video-models): One prompt and one start frame through six video models, side by side.
- Multi-Model Image (https://dgenvisual.com/templates/compare-image-models): One prompt and one source image through six image models, side by side.
- Social post (https://dgenvisual.com/templates/social-post): A post with readable text in the image.
- AI influencer (https://dgenvisual.com/templates/ai-influencer): Make the character, then its continuity sheet: a portrait, a full body of the same person, and the sheet you use for every video after.
- Your first Reel (https://dgenvisual.com/templates/first-reel): Your AI influencer in a 10-second vertical Reel or Story: a hook, a scene and its signature gesture, with sound.
- Your influencer talks (https://dgenvisual.com/templates/talking-ai-influencer): Your AI influencer in a cinematic 8-second clip that speaks with its own voice and lip sync, in a scene that looks filmed.
- Long form video (https://dgenvisual.com/templates/long-form-video): Three clips that run straight into each other, joined on shared keyframes.
- Upscale (https://dgenvisual.com/templates/upscale-image): Make an image sharper and larger, without inventing detail.
- Recast a video (https://dgenvisual.com/templates/recast-video): The same footage with a different person on screen. Motion, cuts and sound stay.
- 360° video (https://dgenvisual.com/templates/360-video): A clip you can look around in, for a headset or a 360 player.
- Extend a scene (https://dgenvisual.com/templates/extend-video): Your clip carries on for a few more seconds, in the same style.
- Sing or talk from a photo (https://dgenvisual.com/templates/talking-photo): One photo and your vocal or voice-over. The face sings or speaks it, in one step.
- Lip sync on your video (https://dgenvisual.com/templates/lip-sync-video): A clip you already have and your vocal. The mouth follows every word, the rest of the shot stays.
- Product in hand (https://dgenvisual.com/templates/product-in-hand): Your product in the hands of a person, then the same shot as a short clip.
- Music video performance (https://dgenvisual.com/templates/music-video): Your artist raps or sings your track, close up and in motion, with the mouth on the vocal.

## For AI assistants

This studio is an MCP server. An assistant can drive it once the user has granted
permission; the generating tools run through the same execution path as the canvas,
so the same pricing and the same checks apply.

- MCP endpoint: https://dgenvisual.com/mcp
- Discovery: https://dgenvisual.com/.well-known/oauth-protected-resource
- Human-readable guide: https://dgenvisual.com/connect

Available tools (26):

- list_presets (read only): The style, camera, lens, light, look, pace and format presets you can pass to generate_image, generate_video and build_script. Their wording is added to the prompt for you.
- build_script: Turns a sequence of shots into a saved flow, without running it. Use this when someone wants something longer than one clip: these models make clips of a few seconds, so 30 seconds is six shots of five. Give one shared "style" describing who, where, what light and which lens, it is repeated in every shot prompt, and without it the shots will not match. Choose kind "storyboard" for one still per shot (roughly a tenth of the price, good for checking) or "video". Returns a flow_id; run it with run_flow.
- list_assets (read only): Lists the user files: everything they uploaded or generated. Use it to see what already exists before making something new.
- get_asset (read only): Returns the details of one file, with a temporary link to view or download it.
- get_credits_balance (read only): The current credit balance. Worth checking before you start something expensive.
- list_flows (read only): The user saved flows, with how many steps each one has.
- get_flow (read only): The make-up of one flow: which steps it contains, which model each step uses, and what is still missing.
- list_templates (read only): The ready-made starting flows a user can begin from.
- create_flow_from_template: Opens one of the ready made templates as a new flow, without running it. Use this when someone asks for something a template already does: it comes with the models chosen, the settings filled in and an example prompt in every step. Get the ids from list_templates. Returns a flow_id; run it with run_flow.
- list_loras (read only): The user own style models (LoRAs). A LoRA only works on the base model it was trained on.
- list_models (read only): The available models with their kind and whether they can work with a LoRA. Use this before choosing a model.
- get_execution_status (read only): The status of a generation or flow run you started earlier. Use the execution_id you were given. If it is still running, this waits up to 20 seconds before it answers, so you rarely need to call it twice for an image.
- list_executions (read only): Your most recent generations, newest first, with their execution_id, status, model and the start of the prompt. Use it to find a run whose execution_id you lost, then poll it with get_execution_status.
- generate_image: Creates an image from a prompt. You get an execution_id back; poll for the status with get_execution_status.
- generate_video: Creates a video from a prompt. Most video models also need a start frame, but text-to-video models do not, check list_models first. Takes minutes: you get an execution_id back immediately and poll with get_execution_status.
- enhance_prompt: Rewrites a short prompt into a fuller description that image models follow more closely.
- run_flow: Runs a saved flow from start to finish. Takes a while: you get an execution_id back immediately.
- create_upload: Step 1 of 2 to add a file to your library: you get an upload_url. PUT the raw file bytes to it with the right Content-Type, then call confirm_upload with the path. PNG, JPEG, WebP, MP4, WebM, MOV, MP3, WAV or M4A, up to 50 MB.
- confirm_upload: Step 2 of 2: confirms a file you uploaded with create_upload and adds it to your library. Returns its asset id, which you can use in generate_image or generate_video.
- list_influencer_options (read only): The settings of dGEN Influencer: 20 groups (type, head shape, proportions, hairstyle, style and more) with the id of every option. Pass the ids to build_influencer_prompts.
- build_influencer_prompts (read only): Step 1 of making a new AI influencer. Pass settings from list_influencer_options, a description, or both, and optionally a reference photo. Returns the exact steps: 1a the portrait, 1b the full body made from the portrait, 2 the character sheet made from both, 3 save_character, and 4 the optional first Reel (generate_video). After step 3 the character exists and works as character_id everywhere. Costs nothing itself.
- list_characters (read only): The characters the user saved in dGEN Influencer, with their images. Pass an id as character_id to generate_image or generate_video.
- get_character (read only): One saved character with its portrait, full body, sheet and profile. Use it before making a video, so you know the signature detail and gesture to write into the prompt.
- save_character: Step 3: saves the character with its portrait, full body and sheet. It then has its own folder in the user library (Characters) and works as character_id in generate_image and generate_video. To add or replace the sheet later, use update_character.
- update_character: Changes a saved character: a new name or profile, or a new sheet, portrait or full body from the library.
- set_character_voice: Gives a saved character a voice, so it speaks its lines itself in Kling 4.0 videos (generate_video with character_id and model kling-v4-reference-to-video; put the line in quotes in the prompt). Pick one of: Aria, Sarah, Laura, Charlotte, Alice, Matilda, Jessica, Lily, Roger, Charlie, George, Callum, River, Liam, Will, Eric, Chris, Brian, Daniel, Bill. Costs 46 credits, once per voice.

## What it does not do

- No model training. You bring a LoRA you trained elsewhere; you do not create it here.
- No free tier without credits: every generation costs something, and it says so up front.
- No assistant access without the user seeing and approving a consent screen first.