AI for Painting Contractors: Can It Show Colours Before You Quote?

Yes, you can show a client colour options from a room photo in minutes, using a paint maker's visualiser or an AI image editor. The preview is reliable for comparing directions, such as warm against cool or pale against dark, and unreliable for the exact shade. Confirm the final colour on a painted sample in the room's own light.

The reason is stacking: the camera, the screen and the room's lighting each shift colour, and a preview adds a software edit on top. Treat the image as a way of getting from ten colours to two, then let a painted patch and a written colour code do the approving.

Follow me on Instagram@sagnikteaches

What a preview can settle, and what needs a painted patch

Before you pick a tool, be clear about the job you are giving it. A preview answers questions of direction and balance. It cannot answer questions about the exact shade, the surface or the light.

Connect on LinkedInSagnik Bhattacharya

That table gives you the working rule: a preview narrows the field, and a patch picks the winner. Everything below is about making the first half quick and the second half unavoidable.

Subscribe on YouTube@sagnikteaches

Three kinds of tool, and what each wants from you

The tools that can recolour a room photo fall into three groups. They differ in where the colours come from, how much of the photo they leave alone, and whether you can attach the result to a quote.

The big advantage of a paint maker's tool is that the colours are real products with codes you can write into a quote. Sherwin-Williams' visualiser page lets you choose a similar scene or add a photo to virtually paint, and its own disclaimer says actual colour may vary from the on-screen version and advises checking a physical chip, card or painted sample before you buy. Behr's visualiser page invites you to upload a photo of your own room, gives you a control for centring and zooming the photo, and says that uploading means agreeing to its website terms and privacy policy. Neither page, as I read it, said how the walls are found, so check that on your own photos.

General editors work differently. OpenAI's help article on image editing says you can upload an existing image and describe the change, or select part of the image first, and it warns that highlights are not always precise and edits may extend beyond the area you selected. Google's Gemini Apps Help describes the same upload-and-ask flow, with examples such as changing a background or replacing an object. Paid individual plans are listed at $20 a month for ChatGPT Plus and $19.99 for Google AI Pro, and image allowances differ by plan, so check yours. Free Gemini downloads were listed at 1K resolution with a visible sparkle watermark when checked in October 2026, and whether a paid plan removes the mark depends on the account.

The third group bolts a preview onto quoting. ServiceM8's pricing page ticks AI image generation on every plan, Free included, and describes it as transforming site photos to help clients visualise the outcome you are quoting. The Free plan caps AI helper uses at 10 a day, and the page does not say whether images count towards that cap or carry a limit of their own, so try it on a free account before you build it into your quoting. QuoteIQ's Before & After AI page says you upload one photo of the property as it is now and get a realistic preview of how it will look once your service is done, which you can attach to an estimate. Its examples lean towards exterior cleaning jobs, the page says the feature is on every plan while its own FAQ places it in an AI Assistant Package, and it says nothing about credits, cost or accuracy, so confirm all three with QuoteIQ. For what painting estimating tools cost and which ones use AI, see painting estimating software and what it costs, and for whole-room renderings that go beyond recolouring, AI remodel renderings from client photos.

Take the room photo like a colour test, not a snapshot

The photo decides more of the result than the tool does. A dim, warm, tilted snapshot will give the same dim, warm, tilted preview in any of them. Before you open a visualiser, shoot the room like this.

• Shoot in daylight with the lamps off and no flash, and note the time and which way the window faces.
• Stand in a doorway or corner so the whole wall, the ceiling line and the skirting boards (baseboards) are in frame, with the wall edges straight.
• Hold the phone level at chest height. A tilted camera leans the walls, and software then struggles to find their edges.
• Tap on the wall to lock exposure, and switch off filters, beautify modes and night modes if your phone has them.
• Put a sheet of plain white paper and the current colour sample card in one corner of the shot, or take a second photo with them in it. They give you something to compare against later.
• Take one wide photo plus one photo per wall in play, and reuse the same photos for every option so the comparison is fair.
• Clear the clutter that confuses an editor: stepladders, tape, bright dust sheets near the wall edge.
• Crop out family photos, papers and faces before you upload anywhere, and tell the client you are using an online tool.

The same set of photos can serve the quote itself, which is covered in how painters and decorators quote from photos with AI.

A prompt for an AI editor, and five things to check in what comes back

If you use a general editor, the instruction does most of the work. Say which walls, what stays, and what the colour is. If the paint maker's page lists an on-screen (RGB or hex) value for the colour, paste that in as well.

Repaint only the walls in this photo in the colour below. Leave the ceiling, skirting boards, door and window frames, doors, floor, furniture, curtains and light fittings exactly as they are. Keep the existing shadows and direction of light. Do not add, remove or move anything.
Colour: [colour name from the paint maker], on-screen value [hex or RGB if listed].
Walls to repaint: the wall facing the camera and the left-hand wall.
Return one image at the same size as the original.

A vague instruction invites changes you did not ask for. Here is how the same request reads before and after.

Run the edit, then run a five-point check before anyone else sees the picture.

1. Edges. Zoom in on the ceiling line and skirting boards. Colour should stop where the wall stops, with no halo or bleed onto trim.
2. Untouched items. Flip between the original and the edit. Ceiling, doors, floor, furniture, switch plates and curtains should look identical.
3. Light. Shadows and highlights should still fall as they did. A flat, even fill looks like a sticker.
4. Details. Sockets, radiators, pictures and textured plaster should survive.
5. Shade. Hold the sample card against the screen and judge it as "same", "close" or "visibly different".

If the ceiling turned pale pink as well, repeat the request with "the ceiling stays white" stated twice and use the selection tool, where there is one, to mark the walls only. If it still bleeds, that is a result for your test log, not a reason to keep fixing the picture by hand.

A 30-minute test that scores each tool on your own photos

Vendor pages show best-case demos, so the only evidence that matters is how a tool treats your photos, in your clients' kind of rooms. Run this once before you show a preview to anyone.

1. Pick a room where you already know the answer: one you painted, with a dried sample, a tin or a colour card to hand.
2. Photograph the room as set out above. Photograph the real sample on the wall in the same light, next to the white sheet.
3. Run the same recolour in each tool you are considering, for example one paint-maker visualiser, one AI editor and a quoting-app preview if you have one. Time each from upload to finished image.
4. View every result on three screens: your phone, a laptop, and a phone or tablet belonging to someone else.
5. Score each tool from 1 to 5 on edges, untouched items, light and shadow, ease of sharing with a client, and note "same", "close" or "off" for shade on each screen.
6. Set the pass mark before you look: for example at least 4 on edges and on untouched items, and "close" or better on shade on two of the three screens.

The scorecard below is made up to show how to fill the sheet in. The scores describe an imagined test and say nothing about any named product.

Read the imagined result like this. The maker's tool passes the untouched-items check and gives you a real code, so it becomes the one you use for choosing a colour with the client. The editor has cleaner edges but needed a re-run and carries no code, so it is kept for the occasional feature-wall idea. The quoting-app preview fails two checks and is not shown to clients, whatever its convenience. Your own result will differ, and the point of the test is that it is yours.

One living room, two colours and a $173 surprise

Here is the sum that the preview cannot do for you. A painter quotes a living room 4.2 m by 5.0 m with 2.4 m walls, one door and one window. The client currently has deep teal walls, previews three pale options on their own photo and picks a warm cream. Everything below is illustrative: use the coverage printed on your tin, your own labour rate and your own prices.

The wall area is the perimeter, 18.4 m, times the height, 2.4 m, which gives 44.2 m². Take off 2.0 m² for the door and 2.5 m² for the window and you have about 39.7 m² of wall. Two finish coats make 79.3 m² of painting.

The preview showed cream replacing teal as a clean swap. It showed nothing about what lies underneath, so the extra coat, $173 or about 50% more, appears only if someone looks at the wall. If a strong colour went on instead, a third finish coat would add roughly another 2.0 hours, $109 of labour, and about 3.3 litres of paint on the same assumptions. Prep, masking, ceilings, trim and overheads are not in these figures and belong in yours.

The habit that follows is simple. Keep a photo of the existing colour next to the preview in the job file, and let the quote say which coats it assumes. The scope, prep and coat wording is drafted in the painting contractor proposal template.

From three previews to one signed-off patch

Show the client no more than three previews, and label each "indicative only" with its colour name and code. Then paint patches, because a patch carries the sheen, the texture and the room's own light, which a screen cannot.

• Paint a patch of roughly 50 cm square or larger, or use a large sample board, with the same number of coats you will use on the job.
• Put one on the brightest wall and one on the darkest, not in a corner where the other walls tint it.
• Ask the client to look in morning light and again in the evening under their own lamps.
• Ask for the answer in writing, in a message that names the colour, the finish and the coats.

A message you can adapt:

Hi [first name], thanks for choosing. To confirm before I order paint:
Room and walls: [for example, living room, all four walls]
Colour: [colour name and code], [brand], [finish]
Coats: [number], with [primer, if needed]
Patch: painted on [wall] on [date]; you have seen it in daylight and in the evening.
The previews I sent are a guide only. The painted patch decides the colour. If you are happy, reply "Approved" and I will order the paint.

Filled in for the living room above, the details read: all four walls, warm cream, the maker's code, matt, two finish coats over one coat of high-hide primer, patch on the window wall viewed on Monday evening and Tuesday morning. If the client then reports that the evening patch looked yellower than the preview, that is the message you want to receive before the paint is on the walls, and the answer might be the cooler cream, on the same quote. The colour-approval clause and change-order wording are covered in the painting contract template tutorial.

Which jobs get a preview, and which get a patch only

Not every job needs the full routine. Match the effort to the cost of getting the colour wrong and to how settled the client already is.

The decision rule fits in two lines. If the client has already chosen, skip the preview and go straight to a patch. If the client has not chosen, preview down to two or three, then patch.

Where the preview sits in the written quote

Attach the preview images to the quote with a line underneath: "Illustrative only. Final colour confirmed on a painted sample." Then write the real specification in the quote itself, area by area: colour name and code, brand, finish, number of coats, and any primer or stain-blocking step. Decide whether sample pots and patch time are included or charged, and say which. A quote that names the colour and the coats gives a client nothing to argue about later, and the preview becomes what it should be, a way into the conversation rather than the contract.

Questions painters ask about colour previews

Can ChatGPT or Gemini repaint a room photo accurately?

They can recolour walls convincingly, but accurate is the wrong test. OpenAI's help page warns that selections are not always precise and edits may extend beyond the selected area. A colour name typed into an editor is not a paint code either, so give it the maker's on-screen value if one is listed, and confirm the shade on a painted patch.

How many previews should I send a client?

Send up to three, each labelled indicative only and named with its colour code. More than three tends to restart the decision rather than finish it. Keep the originals on the job file with the date, so you can show exactly what the client saw when they approved, and say plainly that the painted patch is what decides.

What happens to the client photos I upload to a visualiser?

That is a question for each tool's terms. Behr's visualiser page, for example, says that uploading a photo means agreeing to its website terms and privacy policy, and says no more on that page. Read the policy first, crop out personal items such as family photos and papers, and tell the client you are using an online tool.

Can a preview show matt, satin or gloss?

No. A recolour changes hue and lightness, while sheen depends on the surface, the angle of the light and the paint itself. Name the finish in the quote and show it on a sample board or a patch. Satin on a wall with flaws shows every one that matt would hide, so the finish deserves its own line in the conversation.

Further reads

• AI for a Painting Business: Snag Lists and Final Walkthrough Sign-Off — Snag lists and a final walkthrough sign-off once the colour is up.
• Following Up Silent Quotes With AI: A Decorator's Playbook — What to send when a client goes quiet after your quote.
• Can AI Give Customers a Ballpark Price for Decorating Work? — Whether AI can give a ballpark price before a visit.
• Jobber for Painters: Estimates, Deposits and the AI Receptionist — Estimates, deposits and the AI receptionist in Jobber.
• Shutters Visualiser: Show Customers the Look With AI Before You Quote — The same preview idea for a different trade, with its own limits.
• AI Room Visualisation From Client Photos: A Designer's Workflow — A designer's workflow for room visualisation from client photos.
• Painting Invoice Template: Deposits, Progress Billing and Chasing With AI — Deposits, progress billing and chasing, once the job starts.
• Paint Store Marketing: Colour Advice Posts and Trade Offers With AI — Colour-advice posts built from your own chart, a coverage sum, honest wording on photo matching and a trade-account email with the margin maths shown.
• AI Construction Design Software: What Renovation Contractors Can Draw — A renovation contractor's guide to AI design software: scan-to-plan, layout options and mock-ups compared, with the points where a professional must take over.
• Painting Contractor Marketing: Seasonal Campaigns With AI Content — A seasonal marketing calendar for painters built from your own capacity gaps, with AI drafting Business Profile posts, job captions and customer emails.
• Can AI Concept Images Help Win a Home Renovation Proposal? — How a renovation firm can use ChatGPT or Gemini concept images from a site photo in a proposal, with a not-a-design disclaimer and scope control.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: Sherwin-Williams and Behr colour visualiser pages; OpenAI help article on editing images in ChatGPT; Google Gemini Apps Help (as shown in search results); ServiceM8 pricing page; QuoteIQ Before and After AI feature page; ChatGPT and Google AI plan pages.