Website review
This review may include AI-generated content and is not individually verified. Domain, certificate and other facts come from the observed data on this page.
What is ProsePainter 2?
PROSEPAINTER 2 is a browser-based digital painting tool that uses AI generation as a brush rather than a prompt box. The page's own framing is "Paint - don't prompt": instead of typing a description and waiting for a finished image, you paint on a canvas and steer the AI with styles, reference images and guides. It is aimed at artists who want generative help inside a normal painting workflow, not at people looking for a one-shot text-to-image generator.
What it actually offers
- Style sliders and ingredients. You build a look by combining style "ingredients" on a slider interface, optionally adding custom prompts, and can save and share the combinations as reusable brushes.
- Style by image reference. You can use one or more images as a style reference, including a crop of your own work-in-progress, so a half-finished piece can define the look of what comes next. Styles can be collected into a library.
- Text prompts and AI strength. Prompts guide generation, and an AI-strength control sets how far the output drifts from your own input.
- ControlNet guides. Edge and depth maps let you constrain composition precisely, which matters when the AI would otherwise reinterpret your drawing.
- Canvas growth. You can expand the canvas in any direction up to 5k and upscale for more detail, plus mirroring options and style randomisation for exploring ideas.
Who it suits, and the trade-off
It fits illustrators and concept artists who already paint and want AI to extend, restyle or detail their own marks. The cost of that control is setup: you get the most out of it by preparing reference images and guides, so someone wanting a quick text-to-image result will find it slower than a plain generator.
A concrete test: open the app, paint a rough 10-minute sketch, then try to restyle it using a crop of that same sketch as the style reference at moderate AI strength. If the result keeps your shapes while changing the surface, the workflow fits you.
For comparison, Krea and Leonardo.Ai also combine canvas work with AI generation, while Adobe Photoshop offers generative fill inside a conventional editor.
How do I create and save a custom style brush in ProsePainter 2?
Create the brush on the canvas with the style controls, then save the combination from those same controls. ProsePainter 2 is built around painting rather than prompting: you build a "brush" from style ingredients, refine it, and keep the results you like.
Building the brush
- Start with style ingredients. Use the style slider interface to blend any number of ingredients until the look matches what you want.
- Add a text prompt if you need direction. Prompts are optional guidance on top of the style, not the main input.
- Borrow a style from images. Feed in one or more reference images to capture a medium or aesthetic; mixing several produces hybrid styles. You can also grab a section of your work-in-progress to lock in a look you have already achieved.
- Tune the balance. The AI strength setting decides how much your own input versus the AI's interpretation shapes the result, so adjust it before you commit to saving.
Saving and reusing
Use the save option in the style controls to store the combination, then share it if you want others to paint with it. Saved brushes accumulate into a personal library you can return to, which is the practical payoff: a consistent look across sessions instead of rebuilding it each time.
A workable routine
- Rough in a small area of your canvas.
- Blend two or three ingredients and nudge AI strength until the texture reads correctly.
- Capture that section as a reference to anchor the style.
- Save the combination and name it by purpose ("soft gouache," "ink wash"), not by ingredient.
- Randomize styles occasionally to find new starting points, then save anything worth keeping.
One trade-off to expect: piling on ingredients and image references gives more distinctive brushes but makes results harder to reproduce and slower to tune. Two or three well-chosen ingredients usually beat a long list. For artists who think in marks and texture rather than sentences, this workflow sits closer to traditional brush-making than to prompt engineering—see PROSEPAINTER 2 for the current interface.
How can I use an existing image as a style reference to build a brush?
Use the Style by Image Reference feature: upload one or more images, and ProsePainter turns them into a brush that captures the medium or aesthetic you point it at, rather than copying the subject. You can then paint with that brush on your own canvas.
A practical workflow
- Pick a clean reference. Choose an image whose look you want—texture, color palette, lighting—not one whose composition you want to reproduce. A photo of oil-painted wood grain works better than a busy scene with many competing elements.
- Mix images for hybrid styles. Combine two or three references to blend qualities (e.g., one for color, one for texture). The page notes you can mix several images into a single hybrid style.
- Capture from your work-in-progress. If you like how a section of your current painting is turning out, grab that region as a reference to lock in the style and keep painting consistently.
- Refine with prompts and strength. Add text prompts to steer the AI, and adjust AI strength to balance your own brushwork against the AI's interpretation.
- Save and reuse. Build a library of favorite styles so you can return to a brush across sessions.
When to use it versus other controls
| Goal | Best tool |
|---|---|
| Match a medium or aesthetic from a photo | Style by Image Reference |
| Precise composition or edge/depth steering | ControlNet guides |
| Quick exploration of looks | Style slider / randomize |
| Consistency across a large piece | Capture a section of your work-in-progress |
Next step: start with one strong reference, generate the brush, paint a small test patch, then add a second image only if the first feels too narrow. If you want to compare approaches, see PROSEPAINTER 2.
How do ControlNet guides like edge and depth maps change what the AI generates?
ControlNet guides let you steer where the AI puts structure, not just what it depicts. Instead of relying on a text prompt alone, you feed the model an extra map—edges, depth, and similar—that it treats as a spatial constraint. In ProsePainter 2 this is presented as a way to "steer your generations with precision" alongside text prompts and an adjustable AI-strength slider PROSEPAINTER 2. The practical effect: the AI keeps its output aligned to the geometry in your guide while filling in style, texture and color from your prompt or reference images.
What each guide tends to do
- Edge maps (outlines/line art): preserve contours and silhouettes. Useful when you have a sketch or a photo's linework and want the AI to render it in a new medium without moving the shapes. Expect crisp boundaries but little sense of volume.
- Depth maps (near/far shading): preserve spatial layout and volume. Useful for keeping foreground/background relationships and perspective when you restyle a scene. Expect believable 3D structure, with edges allowed to soften.
A quick comparison:
| Guide | Constrains | Best for | Main trade-off |
|---|---|---|---|
| Edge | Outlines, boundaries | Line art, logos, hard-surface objects, tracing a sketch | Can look flat; interior shading is left to the AI |
| Depth | Distance and volume | Landscapes, interiors, figures in space | Fine detail and crisp edges may drift |
How this plays out in practice
Say you've drawn a rough building sketch and want a painterly version. An edge guide keeps the rooflines and window frames where you drew them while the style slider and prompt supply brushwork. Swap to a depth guide and the AI will respect that the building recedes into the distance, but it may reinterpret your exact window shapes. Because ProsePainter also offers image-reference styles and an AI-strength control, the guide sets structure while those settings set look and how much freedom the model has—lower AI strength keeps you closer to the guide; higher strength lets the model improvise more.
Decision criterion: choose edge guides when shape accuracy matters most (sketches, product forms, typography-adjacent work); choose depth guides when spatial believability matters most (scenes, environments, figures). If you need both, start with the guide that protects what you'd least want the AI to change, then adjust AI strength before adding more constraints.
How do I balance text prompts and AI strength to keep control over my painting?
Use text prompts to say what you want and AI strength to decide how much the model is allowed to change your painting. Prompts set direction; strength sets permission. In PROSEPAINTER 2, the page describes text prompts as guidance for generations and an adjustable AI strength that balances your own input against AI creativity—so the practical goal is to keep strength low enough that your marks still read, and specific enough in the prompt that the model doesn't fill gaps with its own ideas.
A working balance
- Start low, raise gradually. Begin with strength near the lower end and nudge it up in small steps until the AI contributes what you want. It's easier to add than to recover a painting that has been overpainted.
- Make prompts about the gap, not the whole picture. Describe only what's missing ("warm rim light on the left edge," "looser brush texture in the background") rather than restating the entire scene. Broad prompts plus high strength is where control is usually lost.
- Constrain before you prompt. Use edge or depth guides to fix composition, then let prompts handle surface qualities like color, medium and mood.
- Lock in what works. Capture a section of your work-in-progress as a style reference so later generations inherit the look you've already achieved instead of drifting.
Quick comparison
| Situation | Prompt approach | AI strength |
|---|---|---|
| Blocking in composition | Minimal; rely on guides | Low |
| Refining texture or lighting | Specific, localized | Low to medium |
| Exploring alternatives | Broad, varied | Medium |
| Generating over a rough sketch | Descriptive | Medium to high |
| Protecting detailed linework | Very specific | Low |
Whose observation is whose
The page's own claims are about features: prompt guidance, adjustable strength, style sliders, image-reference brushes, ControlNet-style guides and canvas expansion. The judgment that low strength plus narrow prompts preserves control is my practical analysis, not a stated product rule—treat the table as a starting heuristic to test on your own canvas.
Next step: take one small area of a current painting, write a single-sentence prompt naming only the change you want, and run it at low strength. If your brushwork survives, raise strength one step; if it disappears, lower it and make the prompt more specific. That single test tells you more about your personal balance point than any preset.
Can I expand or upscale my canvas without losing detail as my artwork grows?
Yes. PROSEPAINTER 2 is built around exactly that workflow: you can expand your canvas in any direction up to 5k and upscale your artwork for added detail, so a piece can keep growing instead of being locked to its starting size. The page also describes seamless cropping and expansion as the artwork evolves, which matters if you tend to compose loosely and then extend the frame later.
What this means in practice
- Starting small is fine. Sketch a tight composition, then extend outward when you decide the scene needs more sky, foreground or negative space.
- Expansion is directional. You can add room on one side rather than resizing the whole image, which suits fixing a cramped crop.
- Upscaling is separate from expanding. Expanding adds new area; upscaling increases resolution for finer detail. Treat them as two different steps in your workflow.
A concrete scenario
Suppose you paint a portrait that fills the frame, then decide it needs a wider environment. Expand to the left and right first, block in the new area at your working resolution, and only upscale once the composition is settled. Upscaling early tends to lock in decisions you may still want to change.
Trade-offs to weigh
| Approach | Best for | Watch out for |
|---|---|---|
| Expand first, upscale later | Compositions still in flux | More rework if you change your mind after upscaling |
| Upscale first, expand later | Detail-heavy focal areas you want to refine early | Larger files and slower iteration while exploring |
The page's style tools, including image-reference brushes and ControlNet guides such as edge and depth maps, can help keep newly added areas consistent with the existing painting. As a practical next step, test expansion on a single throwaway canvas before committing a finished piece to it, so you can see how much manual blending the new regions need.
User reviews (0)