How skills and nodes work in Piccaso

This tutorial builds two skills in the Piccaso studio. The first turns a product photo into a clean studio packshot. The second puts the same product on a model in front of a background you choose. Each step adds one idea, so by the end you will know what every node does and how a skill becomes a form your team can run.

You need a Piccaso workspace and one product photo. We use a tweed cap.

What a skill is

A skill is a saved image workflow. You build it once, and then anyone on your team runs it by filling in a short form.

Every skill has two views, and you switch between them with the Editor and Run tabs at the top of the page:

• Editor is a canvas of nodes. Each node is one step, such as "take an upload" or "generate an image". The lines between nodes decide which images flow into which step.
• Run is the form that Piccaso builds from those nodes. The person running the skill never sees the canvas. They see one row per input, the model choice, and a Start Generation button.

So when you add a node in the editor, you are also adding a question to the form.

Part 1. A studio packshot skill

Step 1. Create the skill

Open Skills in the sidebar and choose the Custom tab. Click New Skill, then Image.

The first time you do this, Piccaso offers a short guided tour. You can take it or skip it. This tutorial covers the same ground.

To name the skill, click Untitled 1 at the top of the editor and type a new name. We call ours "Studio packshot". The builder saves every change on its own, so there is no save button.

Step 2. Read the starter graph

A new image skill already holds three nodes, connected left to right.

• Media Upload is where the person running the skill adds their product photos.
• Image Generator runs an image model with your prompt. Everything wired into its left side is available to that prompt.
• Output marks which generator's images are the results of a run. It has no settings.

This graph is already a working skill. It only needs a prompt.

Step 3. Write the prompt

Click the Image Generator node. A panel opens on the right with a Base Prompt box.

Type the start of your prompt, then type @. A menu lists the nodes connected to the generator. Pick @Media Upload.

The mention is how the model knows which image is the product. When the skill runs, Piccaso sends the uploaded photos to the model under the name Media Upload, and your prompt points at them by that name. The menu only lists nodes that are wired into this generator, so connect a node before you try to mention it.

Here is the full prompt we use:

A clean studio packshot of @Media Upload on a warm off-white background. Soft daylight from the left and a gentle shadow under the product. Keep the product exactly as it looks in the photo.

Write the prompt the way you would brief a photographer. Name the setting, the light, and what must not change.

Step 4. Decide what the person running it can change

Each node's settings decide which controls show up on the run form.

Click Media Upload to see what it accepts.

• Allowed File Types sets which formats an upload can be: PNG, JPG, WebP and AVIF.
• Allow multiple uploads and Max files set how many photos one run takes. The default is up to 6.
• Allow product description adds a text box where the person can describe the product in their own words.

Now click Image Generator again and open Advanced Settings.

• Allow additional context adds an Additional Context box to the form. Its text is added on top of your prompt for that run.
• Allow output file type selection, Allow aspect ratio selection and Allow resolution selection each add a picker to the form. Turn one off and you choose a fixed default instead. Step 9 does this.
• Allow override gives this generator its own tab on the run form, so the person running it can pick a different model and settings for this step.

For the packshot we leave the defaults.

Step 5. Run it

Click the Run tab. This is the form your team sees.

Click the Media Upload row and add your product photo. Earlier uploads stay under Previous Uploads, so you only upload a product once. Click Continue to return to the form.

The form has one row per input node, then:

• Additional Context, because that switch is on.
• Provider, the model connection your workspace runs on, and a choice of models from that provider.
• Output Configuration, with the file type, aspect ratio, resolution and how many images to make.

Click Start Generation. Piccaso opens the run page and shows each image as it finishes. Here is what our run made from the cap photo.

From there, Re-Run runs the same inputs again and New Run takes you back to an empty form. Every image a skill makes also appears in Outputs in the sidebar, and on the skill's own outputs page.

That is a complete skill. It has one input, one prompt and one result. The rest of this tutorial adds inputs.

Part 2. An on-model shot

The packshot skill puts the product on a plain background. Now we want the cap worn by a model, in front of a backdrop that the person running the skill picks. That needs two more inputs. Create a new image skill and name it "On-model shot".

Step 6. Add a Character and a Background

Click Add Node in the top right of the canvas. The panel lists every node you can add.

The orange nodes are asset nodes. Each one adds a picker to the run form for one kind of reference image:

• Character, the person or model.
• Background, the scene behind the subject.
• Pose, the body position.
• Vibe, a mood or style reference.
• Texture, a material or surface.

The other nodes are:

• Image Generator, a second generator step.
• Video Generator, which makes a short video from a prompt and an optional image.
• Attachment, a file you pin to the skill, such as a logo. The person running the skill cannot see or change it.
• Inspiration, one reference image for overall visual direction.

Click Character, then open Add Node again and click Background.

Step 7. Connect the nodes

A new node does nothing until it is connected. Each node has a small dot on its right edge. Drag from the dot on Character to the dot on the left edge of Image Generator.

Do the same for Background. Then click Tidy layout, the last button in the bottom-left corner of the canvas, to line the nodes up.

Three inputs now feed one generator. The generator still feeds Output, so its images are the results.

Step 8. Mention every input in the prompt

Click Image Generator and type @ in Base Prompt. The menu now lists all three connected nodes.

Mention each one where it belongs in the sentence:

A lifestyle photo of @Character wearing @Media Upload, head and shoulders, in front of @Background. Natural skin texture and soft window light. Keep the product exactly as it looks in its photo.

Mention every input you connect. A connected node that the prompt never mentions still sends its image to the model, but the model has no instruction for what to do with it.

Step 9. Lock what should stay the same

Every on-model shot from this skill should fit the same 4:5 product page slot. So we take the aspect ratio choice away from the run form.

Open Advanced Settings on the generator and turn off Allow aspect ratio selection. A Default Aspect Ratio field appears. Set it to 4:5.

Every run of this skill now makes 4:5 images, whoever runs it. Use the same approach for anything that has to match across a catalogue, such as file type or resolution.

Step 10. Run the on-model skill

Switch to Run. The form now has three rows, one for each input node.

Pick the cap under Media Upload and click Continue. The form moves on to Character. Upload a photo of your model, or pick one from My Characters, which reads from the character folders in your Assets library.

Click Continue, and do the same for Background.

There are two changes from Part 1. The form has a row for each node you added, and the aspect ratio picker is gone because you locked it.

Character and Background say "optional" until something is picked. Asset nodes can be skipped by default. To make one required, turn off Skippable in that node's Advanced Settings.

Click Start Generation to make the shot. Here is ours, at the locked 4:5 ratio.

Every node at a glance

Where to go next

• Try another asset node. Add a Pose node to the on-model skill, connect it, and mention it in the prompt.
• Run many products at once. The More actions menu, the three dots at the top right of the editor, has Run Batch.
• Undo a change. The clock button next to that menu opens Version history, where you can see earlier versions of the skill and restore one.