Need the full Vibart workflow?
Open the main Vibart site to compare models, see pricing, and start your project inside the full canvas workflow.
What is Vibart?
Vibart is a “vibe design” toolkit for creators. You can generate multiple visual directions with AI (images and videos), place them on a canvas to compare and iterate, and publish your results to the community.
Instead of “one prompt, one image”, Vibart is built around a workflow you can actually ship with:
Generate → Place & compare → Edit locally → Export / Publish
The core idea: direction-first + canvas-first
- Direction-first: generate multiple options and select the vibe that works.
- Canvas-first: treat outputs as editable layers, not one-off images.
1) AI image generation: models, resolution, and aspect ratio
Vibart supports multiple models across different needs:
- Fast generation (great for exploration)
- Highest quality (best for finalists)
- Image editing models (best for targeted changes)
In the generator you can also pick common settings like:
- Resolution (e.g. 1K / 2K)
- Aspect ratio (e.g. 1:1, 16:9, 9:16)
2) AI video generation: use a referenced image as the first frame
When your prompt clearly asks for a video, Vibart routes the request to video generation.
- If you reference images in chat, Vibart can pass the first reference as a start frame
- The exposed controls focus on the common set: duration + aspect ratio
3) Canvas editor: iterate like a moodboard
The canvas turns AI outputs into a real iteration space:
- Drag, scale, arrange, and compare
- Smart non-overlapping placement for new assets
- Generation placeholders: a placeholder is inserted during generation, and replaced when done
- Video layers are playable (performance-friendly: play on hover)
4) Mark + Quick Edit: one click to target a local edit
Mark is a fast path for targeted editing:
- Enable Mark mode and click an image to place a blue marker
- The marker becomes a chip/tag in the chat input
- Quick Edit lets you modify the marked area while treating the marker as UI guidance (it should not appear in the final output)
Great for:
- Faces/accessories/local details
- Replacing a local object
- Small style adjustments without destroying the whole image
5) Image tools: Upscale and Remove Background
For delivery-grade assets, two utilities are especially useful:
- Upscale: increase resolution and keep the on-canvas display size in sync
- Remove BG: generate a new cutout image with transparent background for compositing
6) Mobile experience: album mode + bottom chat sheet
Mobile has a dedicated UX:
- Browse images/videos in an album-like gallery
- Chat runs in a bottom sheet with native-like drag and snapping
- During generation, the gallery shows skeleton placeholders and replaces them on completion
7) Community: publish, browse, and reuse
Vibart supports a community workflow:
- Browse public posts and reuse inspiration
- Publish a single output or a full conversation (preview prompts + outputs)
8) Credits: pay as you go (can go negative)
Vibart displays usage as Credits (no subscription).
- Image generation: 6 Credits / run
- Video generation: 14 Credits / run
- Some tools (e.g. Upscale / Remove BG) are billed separately
If Credits are already <= 0, the app will prompt you to top up before generating. If a multi-image run brings your balance below zero, the negative balance is preserved until you top up.
9) Who is Vibart for?
- Designers and creators who need fast direction finding
- Teams who want AI outputs as usable assets, not demos
- Anyone who wants a workflow (generate → edit → ship), not just prompt experiments
FAQ
Q: Why the emphasis on a canvas?
A: Shipping usually needs real layout, alignment, and editable text. AI helps with direction and assets; the canvas helps you compose and polish.
Q: Can I use Vibart just to generate a few images quickly?
A: Yes. Use a fast model to generate 8–12 options, then refine 2–3 finalists with a quality model. It’s usually faster and cheaper in Credits.
Q: Text in images still looks wrong—what should I do?
A: That’s common. Generate backgrounds/layout hints, then add real text as editable layers on the canvas.