Turn a sketch into a photo with AI

Your line drawing becomes a studio photograph, then a second finish, then the video that takes it from paper to object. The full workflow, no account needed.

Interactive demo

Real workflow, real results. Pan and zoom freely.

Use this workflow

The real workflow: the pen sketch of a chair becomes an oak catalogue render, then the same chair in black steel, and the video replays the move from paper to object.

At a glance

What the workflow produces: 2 images in 1:1 · 1 video of 5s in 1080p, 1:1
Canvas structure: 8 nodes wired by 7 connections
Models wired in: Flux Kontext, Kling V3
Cost of one full run: about 113 credits, roughly $1.36

The sketch enters the Media node, the Image node returns a studio render. Everything hangs on respecting the line: a model left free will understand the drawing and redesign its own chair, which gives a lovely picture and a useless tool. The designer wants THEIR object, same proportions, same backrest, same legs.

So the instruction names the parts one by one and leaves open only what the sketch does not say: material, light and backdrop. Flux Kontext edits your drawing, it does not take inspiration from it. The second node starts from the render and not from the sketch: that is what guarantees the two finishes show the same object from the same angle, and not two neighbouring readings.

The video is the extra that stops the scroll: the sketch fills with material, the paper fades, the chair gains volume and becomes a photographed object, in five seconds generated by Kling V3. The full run, two renders and the video, costs just over a dollar.

How it works

1
Photograph your sketch
A clean line drawing, flat, in daylight, with no shadow cast by your hand. The Media node takes a phone photo: what matters is how sharp the line is, not the resolution.
2
Describe the material, not the shape
The shape is in the drawing. The instruction adds what the sketch does not say: solid oak, woven cord, brushed steel, and the studio light. Keep the part that locks the geometry, that is what makes it faithful.
3
Generate the second finish
The second Image node starts from the render: same object, same shot, different materials. Enough to present two finishes of one piece without shooting again.
4
Run the transformation video
The sketch becomes the first frame, the render the last. Five seconds where the object rises off the paper, a format that works on its own in social feeds as much as in a client presentation.

Variations and use cases

The same canvas, tuned for different needs: each variation is two or three lines of prompt away.

Product sketch to catalogue photo

The direct use case: a product drawing becomes a presentable packshot before a single unit exists. Useful for a funding deck, a pre-order or a trade show, as long as you say the object is not manufactured yet.

Two finishes of the same object

The second node starts from the render and not from the sketch, so both versions show the same object from the same angle. That is what lets you build a credible finish board: oak and black steel, two images, one geometry.

Architectural sketches and interior perspectives

A line perspective becomes an atmosphere render. Name in the instruction what has to hold, vanishing lines, ceiling height, position of the openings, otherwise the model puts a bay window where you had a casement.

Children's drawings and personal illustration

The same workflow makes a child's drawing real, but the instruction changes meaning: you are after fidelity to the spirit of the drawing, not to the geometry. The child's drawing workflow is built for that and gives a better result.

A client presentation board

Sketch, render, variant and video fit on one board and tell a design intention in three images. It is often more convincing than an isolated render, because you can see where the object came from.

Jewellery, lighting, footwear

Nothing ties this workflow to furniture. In the instruction, swap the list of parts to lock, seat and backrest, for the ones your object has: upper and sole, setting and stone, shade and stem. The rest of the workflow does not move.

Common mistakes

A sketch that is too loose

The model follows the drawing, it does not correct it. A line hesitating between two leg positions gives an object that hesitates too. Redo the sketch before generating: five minutes with a pencil beats ten renders.

Describing the shape in the instruction

The shape is already in the drawing, repeating it in words hands the model two sources that contradict each other. The instruction should only carry what the sketch does not say: material, finish, light, backdrop.

A shaded or coloured sketch

A drawing already rendered in volume gives a flattering but less faithful result: the model follows your shadows instead of the geometry. Pure line, even a dry one, gives a better render, and that is exactly the case shown in the demo.

Regenerating the variant from the sketch

Going back to the drawing for the second finish gives you two neighbouring objects instead of one object in two materials. The cable comes off the first render, not the Media node: that is what makes the finish board hold.

Recommended models

Flux Kontext

Kling V3

Seedream 5 Pro

Frequently asked questions

Does the render really respect my proportions?

That is the entire job of the instruction, which explicitly locks the proportions, the seat height, the curve of the backrest, the number and position of the legs. The canvas keeps the sketch next to the render: the comparison takes one glance.

Do I need to be able to draw?

You need a line that commits. A clean sketch, with proportions that hold and a single point of view, gives a faithful render; a hesitant sketch gives a hesitant object, and no instruction makes up for a line the drawing never settled.

Does it work on anything other than furniture?

Yes: products, packaging, a shoe, jewellery, a lamp, an interior plan. Just rewrite the list of parts to lock in the instruction, naming the parts of your object instead of the seat and the backrest.

What does a full run cost?

Just over a dollar for both renders and the five-second video. The price shows on each node before you click, and the two images alone cost a few cents if you do not want the video.

Are my drawings used to train a model?

No. Your media stays in your workspace, it is not handed to any training run. It is the question every designer asks before uploading an unregistered project, and the answer does not change with the plan.

Can the render be used commercially?

Yes, the renders are yours. The caution is on what you feed in: the sketch of an existing protected object stays subject to its author's rights, and the render changes nothing about that.

Can I get several angles of the same object?

Not with this workflow, which keeps the angle of the sketch. For a set of angles, draw the second viewpoint and run that sketch through the Media node, or start from the render with the product photo workflow.

How is this different from a 3D render?

A 3D render starts from a measured model and serves manufacturing; this one starts from a drawing and serves showing. It arrives earlier, in seconds, at the stage where the question is not yet how to produce but how to convince.

Turn a sketch into a photo with AI

Use this workflow

Free account, no card required. The workflow opens pre-filled in your canvas.