AI Painting
How to repaint a 3D model with AI in the Texture Painter: writing prompts, choosing a model and quality, what the AI sees and the depth guide, all-sides (360°) generation with reference images, revealing AI layers, upscaling and credits.
AI painting turns a short description into a new texture on your model. You frame the side you want to change, describe the change, and the AI repaints what it sees. The result comes back as an AI layer that you reveal exactly where you want it. This page covers prompts, models, credits, what the AI sees, all-sides generation, revealing and upscaling.
AI painting needs a paid plan and uses credits. Each model shows its cost in the app, and the Generate button shows exactly what the next generation costs. On the free plan the button reads Upgrade to generate.
Generate an AI layer
Frame the side you want to change
The AI repaints the view inside the corner marks labelled What the AI sees. Turn and zoom the model so the area you want to change fills that square. A straight front or side view gives the cleanest result: hold Alt + Shift while you drag to snap to one.
Describe the change
Type it in the prompt box of Generate with AI. Up to 1,000 characters. The lightbulb (Ideas) offers prompts that work well, the clock (Recent prompts) brings back prompts you used, and the book (Prompt library) opens your saved prompts and templates.
Pick a model and quality
Click the model card to choose a model (see the table below). Below it, pick the quality. The cost on the Generate button updates as you choose.
Generate
Click Generate or press Ctrl+Enter (Cmd+Enter on a Mac). The button shows the step the generation is on, such as AI is painting with the time left, then Sharpening the result or Projecting onto the model. A new layer appears at the top of the list with Generating… and the same step. You can keep working while it runs.
With All sides (360°), the button shows Generating, then Making views sharper (with Sharper views on), Applying view 3 of 9 and Blending views. Each view gets its own layer in a new 360° group. A view's row says Waiting its turn until it is projected onto the model.
Write prompts that work
Describe the surface: colours, materials and finishes. The AI keeps the model's shape and your camera angle, so you don't need to describe them.
- Be specific about colour: matte black, pearl white, warm golden brown.
- Name the material: brushed steel, aged leather, polished marble, weathered stone.
- Add the finish: glossy, matte, metallic, chipped paint.
- Say what to keep when you only change a part: keep everything else the same.
- Avoid layout changes such as moving, adding or removing parts. The AI paints surfaces; it can't change the shape.
| Goal | Example prompt |
|---|---|
| Change the colours | Paint the fur a warm golden brown with darker stripes along the back |
| Change the material | Rusty iron with chipped blue paint |
| Add details | Add glowing blue runes on the forehead and cheeks, keep everything else the same |
| Add wear | Add dirt and scratches on the paws and legs, keep the colours |
| Change the style | Hand-painted stylised wood, warm highlights |
Choose a model
Click the model card (Change) to open the picker. Models are grouped by what they do; type in the search box to find one, and click a model to pick it. Each model shows its typical time, its cost range and whether it reads reference images. GPT Image 2.5 Flare at Standard quality is the default (Start here): a sharp, detailed result in about a minute for 3 credits. Qwen Edit Lightning gives results in seconds for 1 credit, great for trying ideas.
Repaint your view (the main models):
| Model | Typical time | Credits shown in the app | Reference images |
|---|---|---|---|
| Qwen Edit Lightning | Instant (~5s) | 1 | No |
| Nano Banana 2 | Fast (~120s) | 7 to 20 | Yes |
| Nano Banana Pro | Pro (~200s) | 14 to 25 | Yes |
| Qwen 2 Pro | Detail (~80s) | 5 | No |
| GPT Image 2 | ~90s | 3 to 25 | Yes |
| GPT Image 2.5 Flare | ~60s | 3 to 25 | Yes |
| GPT Image 2.5 Sunburst | ~120s | 3 to 25 | Yes |
Clean up: Topaz Refine (1 credit) refines your existing texture and stays close to the original.
Classic: Original Texture AI, Fast Texture AI, Creative Texture AI and Texture AI Remove BG (4 credits each), the presets from earlier versions. They add Creativity and Resemblance sliders under More options.
Quality
Below the model card you pick the quality. The options depend on the model:
| Quality | Output size |
|---|---|
| Draft | 0.5K |
| Standard | 1K |
| High | 2K |
| Ultra | 4K |
Higher quality costs more. For example, the GPT Image models cost 3 credits at Standard, 14 at High and 25 at Ultra. Hover a quality to see its cost.
Start cheap, finish sharp. Try your prompt with Qwen Edit Lightning first. When the idea works, generate the final version with a sharper model or a higher quality.
What the AI sees
The AI receives the square inside the corner marks in the 3D view, as it looks in your current shading. To check it before you spend credits:
- Switch the 2D view to What the AI sees. It follows the model as you turn it and updates by itself when you paint or change anything that affects the picture. With All sides (360°) on, it shows every view at once and the corner marks in the 3D view go away.
- Or use Export › Screenshot (Preview what the AI sees).
Turn off AI capture frame in the view bar if the corner marks get in your way while painting.
Depth guide
In Clay and Normals, the AI only sees the shape of your model, and parts that face the same way look alike. An arm in front of the body, a weapon held in front, a tail or crossed legs can then blend into what is behind them, and the AI paints them as one surface. The Depth guide helps the AI see which parts are in front. It is added only to what the AI gets: your 3D view and your painting never show it.
It can add these marks to the picture:
- Outlines where one part is in front of another. On by default.
- Depth shading: depth shown as brightness. Off by default.
- Contact shadows: soft shadows where parts meet or overlap. Off by default.
In Clay the AI also gets a clearer light that shows the shape better. The AI is told that these marks only show what is in front and are not part of the surface. If it copies an outline anyway, the painter removes it before the result goes on your model.
Set it up
Open More options › What the AI sees. Depth guide shows the shading mode it applies to, and remembers its settings for each mode:
| Shading | Depth guide by default |
|---|---|
| Clay and Normals | On: outlines only, at Strength 40 % and Thickness 4 px |
| Textured and Unlit | Off. When you turn it on: the same outlines |
- Strength sets how strong the outlines and contact shadows are.
- Fine-tune turns each mark on or off and sets the outline Thickness, from 1 to 12 px.
- Reset to defaults brings back the settings above for the current mode.
Depth shading
Turn on Depth shading in Fine-tune to also show depth as brightness, then pick how:
| Style | What it looks like |
|---|---|
| By distance | Parts further away get darker. The part nearest to you keeps its brightness. |
| Halo | Nearer parts get a little lighter, with a soft shadow on the surface just behind them. |
Shading strength sets how strong it is, on its own, separately from Strength. In Normals, Textured and Unlit, shading changes the colours the AI sees, so keep it low there or leave it off.
The 2D view in What the AI sees mode updates while you change them, and Preview the picture shows the picture with and without the depth guide.
The separate depth guide image
Models that read several images (the GPT Image models and Nano Banana) can also get a second picture, Also send a depth guide image: a grey image of the same view where brighter means closer, with lines where one part is in front of another. Your picture can then stay clean while the AI still knows what is in front. It is off by default; turn it on in Fine-tune. It costs nothing extra. With All sides (360°) on, it has the same grid of views as the picture. Models that read only one image, such as Qwen Edit Lightning, get the marks in the picture only; Fine-tune says so.
When to change it
- Turn it up when the AI still merges parts: an arm painted in the body's colours, or a strap that disappears into the clothes. Thicker outlines or Depth shading By distance can help too.
- Turn it down or off when you see faint lines or dark edges in the result, or for simple shapes with no overlapping parts.
- Turn it on in Textured or Unlit when the model's colours make parts hard to tell apart, for example a dark model or parts with the same colour.
All sides at once (360°)
A normal generation repaints only the side you are looking at. Turn on All sides (360°) to texture the whole model in one go: the AI paints it from several views at once, so front, back and sides match, and the views are blended into one seamless result.
Use it for overall changes such as a new material or colour scheme. For a detail on one side, a normal generation is faster and more precise. You can also guide all sides with reference images: see Reference images in 360°.
Set it up
| Option | Choices | What it does |
|---|---|---|
| Views | 4 views (Quick), 6 views (Standard, the default), 9 views (Full) | Quick paints the front, right, back and left, each picture larger, but tops and undersides may stay uncovered. Standard adds the top and the bottom: good for most models. Full goes all around every 60°, plus the top, the bottom and a close-up of the front. Best for characters (cleaner faces) and detailed models. |
| Sharper views | On (default) / off (+1 credit a view, 2 at 8K texture size) | Sharpens each view on its own before it goes on the model, sized for your texture size, for finer detail. Under it, pick the upscaler: Detailed (the default), Crisp or Light, the same as in Upscale mode. |
| Blend views automatically | On (default) / off | Blends the views into one seamless result as soon as they arrive, the front views first. |
| Front of the model | Use my current view as the front, Reset | The front gets the most care and wins where views overlap. Only needed if the model faces sideways or backwards: turn it so its front faces you, then click. |
The diagram next to the views shows where they stand around the model, seen from above, with the front as the big dot. The line at the bottom adds up the cost, for example 9 credits in total: 3 for the AI picture + 6 for sharper views (1 per view). Without Sharper views, all three view choices cost the same.
Write the prompt for the whole model. Describe the colours and materials all around, for example Grey timber wolf with thick layered fur, a darker back, a pale cream belly and muzzle, amber eyes. Only describe surfaces the model already has: the AI doesn't add clothing, hair or props.
How each view is framed
Each view is fitted to your model's outline, so the model fills its picture and the AI has as many pixels as possible to paint it. A little space stays free around it: 5 % of the picture on each side.
To change it, open More options › What the AI sees and move Space around each view (0 to 20 %). The 2D view in What the AI sees mode shows the change right away.
- Less space shows the model larger in every view, for more detail.
- More space leaves the AI room at the edges. Try it if thin parts, such as ears, tails or antennas, come back cut off.
With Name the views on, the names stay off the model: where a name would cover it, that view moves the model aside or shows it a little smaller. Otherwise the name could end up painted into your texture.
Reference images in 360° (experimental)
Add up to 3 images to a 360° generation and tell the AI what each one is for: an art style, materials, colours, the object's own concept art, or a logo. The AI then paints every side to follow them. This is experimental: results vary more than with a prompt alone.
Add the images
With All sides (360°) on, find References in the 360° options and click Add a reference. You can also paste an image, drop a file there, or drag an AI, 360° or decal layer from Layers onto it.
Give each image a role
Click the role next to an image and pick what it is for. Add a short note where it helps, for example for the boots or on the chest.
Say how to use them (optional)
Under How to use the references, write one line about how the images work together, for example Keep the design from the concept art, painted in the style of the other image.
Check what the AI gets and generate
What the AI gets shows the pictures in the order the AI receives them: your views first, then each image with its role. Click Generate.
| Role | What the AI takes from the image | Note |
|---|---|---|
| Style | The art style, rendering and brushwork, for example hand painted, stylized or realistic. Never what the image shows. | Optional, for example soft painterly strokes |
| Materials | Its materials and finishes, such as this leather or this metal | Where, for example for the boots |
| Colours | Only its colour palette | Optional |
| This object | The same object, for example the concept art or photo the model was made from. Every view follows its design, colours and details, the front views most closely. Its pose and background are ignored. | Optional |
| Logo or detail | A logo, emblem or pattern, placed where the note says, on every view that shows that spot | Where, for example on the chest |
Use the original image. If you opened the model from your dashboard and it was made from an image, Use the original image appears next to Add a reference. One click adds that image as This object.
Which model to pick. GPT Image 2.5 Flare follows references best and is the one to start with. For copying an art style, Nano Banana Pro also works well. GPT Image 2, GPT Image 2.5 Sunburst and Nano Banana 2 take references too. Models that read only one image, such as Qwen Edit Lightning, show a note with a button to switch to GPT Image 2.5 Flare. References don't change the cost.
Tips for good results:
- Use 1 to 3 clear images, and give each one a single job.
- Use This object for concept art or a photo of the same model. It keeps the design closest to the original.
- Say where materials and logos go in their note.
- Keep writing your prompt for the whole model; the references add to it.
If the AI changes the layout. The AI has to keep your views exactly where they are. If it changes their layout, nothing is applied and the painter doesn't try again on its own, so no more credits are used. A message suggests fewer or simpler images, or another model.
The 360° group keeps its references and roles, also in a saved session. To try again with them, choose Use these references in the group's ⋯ menu: they come back in Generate with AI, with their roles, your prompt and the number of views.
The result: a 360° group
The views arrive as one group in Layers, named after your prompt (for example 360° · Grey timber wolf…) and marked 360°. Under the name it says how many views it has and whether they are blended, for example 9 views · blended. Each view is its own layer with the share of the model it covers, for example Front right · 360° view · 15 % of the model.
Select the group to show its bar:
| Button | What it does |
|---|---|
| Re-blend (Blend views if the views aren't blended yet) | Blends the views again into one seamless result |
| Blend settings | Fine-tunes the blend (see below) |
| Coverage | Shows which view each part of the model comes from in the 2D view. A number on it counts the areas no view saw. |
| Merge into one layer | Turns the group into one layer with exactly this result, and frees memory. Undo with Ctrl+Z if you change your mind. |
| ⋯ | Show raw views (every view at full strength, to compare), Reset blend (default settings, painted edits cleared), Fill gaps, Use these references (when the group was made with reference images) and Ungroup |
Fine-tune the blend
Open Blend settings on the group's bar. Changes show on the model right away, and your painted edits on the views stay.
- Prefer front (default 60 %): how strongly the front views win, so faces stay clean.
- Seam softness (default 40 %): how wide the transitions between views are.
- Match colours (default 80 %): evens out brightness and colour differences between views.
- Areas no view saw: Show below lets the layers under the group show there; Fill in fills them from the nearest view.
- Reset to defaults brings back the default settings.
Fix a view by painting
Each view is a layer you can paint. Select a view in Layers and paint over the model: the brush shows more of that view where you paint, the eraser shows less of it. Use it to bring in the view that looks best in an area, or to hide one that looks wrong. Re-blending and changing the blend settings keep these edits.
- Hold
Alt(Optionon a Mac) and click a view's eye to show only that view. Do it again to show them all. - The eye of the group hides or shows the whole result.
Check the views and the coverage
The 2D view has two extra modes for a selected 360° group:
- 360° views shows every view in a grid with the share of the model it covers. Hover a view to see where it lands on the model, and click it to select its layer. Right-click a view for Prefer this view or Use less of this view: the blend then gives it more or less room.
- Coverage colours the model by the view each part comes from. Areas no view saw show in magenta, and the legend counts them.
Fix gaps
Some spots, such as the inside of a mouth or the gap between legs, are hidden from every view. When the legend says, for example, 5 areas aren't covered by any view, click Fix gaps. The camera turns to the largest gap and Generate with AI opens with your prompt and All sides off. Click Generate: the new view joins the group and the blend includes it. Repeat for the next gap if you need to. Deep crevices that no camera can see stay uncovered; paint them by hand or choose Fill in in the blend settings.
When a gap sits behind another part, like the body under an arm, the coverage legend links to See through (Experimental). It cuts away the part in the way, and its Find hidden spots list has a Show me button that turns the view to a spot and cuts away what covers it. A generation from there joins the group. All sides (360°) is off while the cut is on.
More options
More options at the bottom of the panel holds settings most people never need. The defaults work well. They come in three groups.
| Group | Option | What it does |
|---|---|---|
| Copy a look from images | Reference images | Up to 3 images whose materials and colours the AI copies. Add them from your computer, or drag an AI, 360° or decal layer from Layers onto them to reuse its picture. Only models that take reference images use them. Marked as experimental. With All sides (360°) on, the references move to the 360° options, where each one gets a role. |
| What the AI may change | Keep the shape and angle | On by default: only materials and colours change. Off: the AI may also change the shape. |
| Start from the original texture | Off by default. The AI works from the model's own texture and ignores your paint. | |
| What the AI sees | Picture size | 1K, 2K (default) or 4K: how detailed the picture of your model is that the AI works from. |
| Background | Dark, White or Green behind the model in that picture. Pick one that stands out from the model's colours. | |
| Show the floor grid | Includes the floor under the model in the picture. | |
| Name the views (360° only) | Writes Front, Back, Left and so on onto each view, so the AI keeps them apart. | |
| Space around each view (360° only) | How much empty space each view keeps around the model, 0 to 20 % (default 5 %). Less space shows the model larger, for more detail. See How each view is framed. | |
| Depth guide | Helps the AI see which parts are in front. On in Clay and Normals, off in Textured and Unlit; remembered per shading mode. See Depth guide. | |
| Preview the picture / How shading matters | Shows the exact picture the AI will get, and how the shading mode changes it. |
What happens after a generation
The result arrives as a new layer named Projection 1 (then Projection 2, and so on), marked AI layer. It is placed on top of the layer list and starts hidden: the model doesn't change yet. The 2D view shows the result on the model so you can judge it.
Why hidden? AI results are rarely perfect everywhere. Starting hidden means you only bring in the parts that look good, and everything else stays as it was.
Reveal or hide an AI layer
With the AI layer selected, the options bar says Reveals the AI layer:
- The brush reveals the AI texture wherever you paint. Your brush colour isn't used on AI layers.
- The eraser hides it again.
- Opacity and Hardness work as usual: soft, low-opacity strokes blend the result in gently.
- Paint colour adds a normal paint layer if you want to paint with your colour instead.
Reveal everything with smooth edges
Open the AI layer's ⋯ menu and choose Apply with smooth edges. A small Apply Layer panel opens over the 2D view, and the 3D view shows a live preview:
- Coverage (30 to 100%, default 85%): how far the result reaches toward the edges of what the AI saw.
- Edge Softness (0.2 to 4.0, default 1.0): how softly it fades out at those edges.
- The eye button compares with and without the result.
Click Apply to reveal the layer, or Cancel.
Upscale an AI layer
For a sharper copy of an AI layer, open its ⋯ menu and choose Upscale. The Upscale Layer dialog offers 4 MP (about 2048 × 2048, 1 credit) or 8 MP (about 2828 × 2828, 2 credits), with Enhance details and Enhance realism. The sharper version arrives as a new layer marked AI · upscaled.
Upscale a blurry texture
When a texture looks blurry but you like its design, sharpen it without writing a prompt. At the top of Generate with AI, switch from Repaint to Upscale, choose what to sharpen, pick how, and click Upscale.
| Choice | What it sharpens |
|---|---|
| This view (default) | What's inside the capture frame, put back on the model with soft edges, like an AI layer. |
| All sides | 4, 6 or 9 views around the model, each sharpened on its own and blended into one result, like All sides (360°). |
| Whole texture | The texture itself, flat: every part of the model at once, including hidden spots, with no seams between views. |
Under the choice, pick how to sharpen:
| Option | What it does | Settings |
|---|---|---|
| Enhance (default) | The AI (GPT Image 2.5 Flare) redraws the view at high resolution with crisp detail: defined eyes, hair strands, fur and fabric, in your texture's own style. The best choice when a texture is very soft. It can change small details, so hide the new layer to compare. Not available for Whole texture. | 2K or 4K redraw: This view starts at 2K, All sides at 4K, because its views share one picture (6 views get about 1570 pixels each at 4K, about 840 at 2K). Sharpen to your texture size (on by default) then brings the redraw up to your texture size with Detailed. |
| Detailed | Rebuilds fine detail and stays true to the picture. The sharpest plain upscale; it can't add much to a very soft texture. | Added detail: low stays closest to the picture; higher invents more fine detail like scratches and grain, and near the top it can change small shapes. |
| Crisp | Clean, sharp edges. It always makes a picture 4 times larger, so it starts from a smaller picture. | None. |
| Light | The lightest touch, up to about 2800 pixels. | Add fine detail (on by default) adds small detail like fur strands and fabric weave. Photo-real look (off by default) pushes towards a photographic look; leave it off for stylised or hand-painted textures. |
The size of the result follows your texture size, and so does the cost. At 4K texture size:
| Choice | Enhance | Detailed | Crisp | Light |
|---|---|---|---|---|
| This view | 16 credits (2K redraw 14 + 2 to sharpen) | 2 credits (1 below 4K, 3 at 8K) | 1 credit | 2 credits (1 below 4K) |
| All sides | 4K redraw 25 + 1 per view to sharpen (31 for 6 views; 2 per view at 8K) | 1 credit per view (2 at 8K) | 1 credit per view | 1 credit per view (2 at 8K) |
| Whole texture | Not available | 4 credits (1 up to 2K, 16 at 8K) | 4 credits | 4 credits |
A 2K redraw costs 14 credits, a 4K one 25. When the redraw is already about your texture size, nothing is added to sharpen it. The Upscale button always shows the total.
The result is a new layer on top (Enhanced view, Upscaled view, a 360° group, or Upscaled). Your other layers stay as they are: hide the new one to compare, or erase it where you prefer the softer look. This view and All sides capture the model in Unlit, so the lighting isn't sharpened into the texture; the view switches to Unlit if needed.
The result has your texture size, so for a texture that is too small for the model, raise the texture size in the top bar first, then upscale. If a step fails, nothing is added and a message tells you.
Retry, cancel and history
- Not happy? Change the prompt, model or view and generate again. Each result is its own layer, so you can compare and delete the ones you don't need.
- Cancel a running generation by clicking the Generate button (it reads Cancel while running), the thumbnail of the layer that is on its way, or Cancel in the 2D view.
- The clock icon at the top of Generate with AI lists the AI layers in this session and the prompts you used. Click one to select it again.
Worked example: hand-painted eyes, then an AI weathering pass
This combines both ways of painting, and shows why layer order matters.
- Paint the eyes by hand. Add a Paint layer and name it Eyes. Point at an eye, press
Fand zoom in. Pick a bright colour, make the brush small with[, raise Hardness and paint the iris. Turn the model and do the other eye. - Frame the body. Press
Shift+F, then snap to a front view withAlt+Shift+ drag. - Generate a weathering pass. Prompt: Add dirt and scratches on the paws and legs, keep the colours. Start with Qwen Edit Lightning.
- Reveal only where you want wear. Select the new AI layer and paint over the paws and legs. Leave the face alone.
- Keep your eyes on top. Drag the Eyes layer above the AI layer, so the AI result never covers your hand-painted detail.
Credits and errors
- The Generate button always shows the cost of the next generation. Each model's cost range is in the picker.
- AI stickers (decals) cost 5 credits; upscaling a layer costs 1 or 2 credits. In Upscale mode the cost depends on the choice and the texture size (see above); the Upscale button always shows it. With All sides (360°), Sharper views adds 1 credit per view (2 at 8K texture size).
- If a generation fails, for example because you are out of credits, an error message explains why, no layer is added, and the 2D view shows what happened. Click the credit balance in the top bar to get more credits.
Related
- Layers: masks, order and merging.
- The 2D view: check results and what the AI sees.
- Tips and troubleshooting
Frequently Asked Questions
New AI layers start hidden so you decide where the result shows. Paint over the model with the AI layer selected, or choose Apply with smooth edges in its layer menu.
The default, GPT Image 2.5 Flare at Standard quality, gives a detailed result for 3 credits in about a minute. To test many prompts quickly, switch to Qwen Edit Lightning: a result in seconds for 1 credit.
In Clay and Normals the AI only sees the shape, and parts that face the same way look alike. Keep the Depth guide on (it is by default in these modes), turn its Strength or the outline Thickness up, or add Depth shading. With a model that reads several images, such as GPT Image 2.5 Flare, you can also send a separate depth image. See Depth guide.
No. AI painting only changes the texture. Prompts that ask to add or move parts won't change the shape.
For changes that should look the same all around the model, such as a new material or colour scheme. For one detail on one side, a normal generation is faster and more precise.
6 views (the default) suits most models: the four sides plus the top and the bottom, so shoulders, roofs and undersides are painted too. 4 views gives each view a larger picture but can leave tops and undersides bare. Choose 9 views for characters and detailed models: it adds the corners and a close-up of the front, so faces come out cleaner and the fewest spots are missed. The AI picture costs the same for all three; Sharper views adds 1 credit per view (2 at 8K texture size).
Each view already fits the model with 5 % of space on each side. For even larger views, open More options › What the AI sees and lower Space around each view, down to 0 %. If thin parts then come back cut off, raise it again. See How each view is framed.
One view probably painted it differently. Open 360° views in the 2D view and hover the views to find it. Then right-click it and choose Use less of this view, or select it in Layers and erase it in that area. Another view takes over there.
Yes. Select the group and click Merge into one layer on its bar. You get one layer with exactly the result you see, and the views are removed. Ungroup in the ⋯ menu keeps the views as separate layers instead.
Yes. Open More options and add up to 3 images under Copy a look from images. To reuse an earlier AI result, drag its layer from Layers onto them. The AI copies their materials and colours. Only models that take reference images use them.
Yes, as an experimental feature. With All sides (360°) on, add up to 3 images under References in the 360° options and pick a role for each: Style, Materials, Colours, This object or Logo or detail. GPT Image 2.5 Flare follows them best. See Reference images in 360°.
Painting tools
Every painting tool in the Texture Painter: brush, eraser, image brush, pattern brush, decals, eyedropper and fill, plus colours and palettes, brush size, opacity and hardness, Paint hidden areas, Edge fade, and See through for areas behind other parts.
Layers
How layers work in the Texture Painter: layer types, adding, renaming, reordering, grouping, merging and deleting layers, visibility, lock, opacity, blend modes, masks on AI layers, colour adjustments, filling unpainted areas, and undo.