Badge Presentation Prompt Builder (DAELab)

What it is and why you'd reach for it

Every other prompt node in the DAELab badge pack is trying to stop the model from having ideas. Lock the text, lock the colours, lock the camera, render it flat and boring because the flat boring version is the one the masks correspond to. This node is the release valve at the end of the pipeline: now that the badge is finished, turn it into a product photograph.

That's a real workflow step, not vanity. Manufacturing documentation wants the flat render; the client deck wants the badge on a soft white sweep with a contact shadow and a highlight running down the rim. Badge Presentation Prompt Builder writes the permission slip - studio key/fill/rim lighting, metal edge highlights, material reflections, contact shadow, ambient occlusion, depth of field, macro framing, slight final perspective - while still forbidding structural change.

How it works

It's the simplest node in the pack and that's the point: one multiline text field, one output. presentation_direction defaults to Premium studio product photograph and is dropped into an Art direction: slot at the end of a fixed instruction block.

The fixed block does three jobs. It declares Image 1 the completed registered badge master. It lists what must survive - design, exact visible text, colours, material regions, outline, internal structure, front-to-back height relationships - and forbids adding or removing structure or changing what the logo means. Then it grants the freedoms, explicitly, including "a slight final perspective". That last one is the interesting clause: after this node, the image is no longer in design coordinates.

Which is why the prompt closes with the sentence that made me trust this pack's author a little: "This image is a presentation result and is not reused as a design-coordinate edit source." They wrote the constraint into the prompt itself, because the obvious mistake here is to take the pretty studio render and try to do one more local edit against it. Your masks won't line up, because the model just added perspective and a macro crop.

Inputs and outputs that matter

Just presentation_direction. Write it like an art director would brief a photographer - surface, angle, light, background: "brushed steel badge resting on warm off-white paper, 45° three-quarter view, soft top-left key with a wide falloff, no props." The default is deliberately bland so it doesn't fight your brand.

Output is prompt, one string, into the GPT-Image node's prompt input. No image sockets, no masks - this is pure text plumbing, which is why it's cheap to experiment with. It costs nothing until you queue.

Install

cd ComfyUI/custom_nodes
git clone https://github.com/TeutonicMepheles/ComfyUI-DAELab-Custom-Nodes-Library

Restart afterwards. Manager install by pack title is equivalent. This node itself is dependency-light (it lives in the badge module, which imports OpenCV for the pixel nodes around it - pip install opencv-python-headless if the node doesn't appear). The generation it feeds is a hosted model: expect prepaid credits per call, and expect your image to leave the machine.

Common issues

• The render comes back with a different badge. Overly clever art direction does this - "floating in a dark void with neon rim light" invites re-interpretation. Keep presentation_direction about light and surface, not about drama.
• Text gets mangled in the studio pass. It's the usual failure of instruction editing on dense small type. Do the presentation stage last, check the copy, and re-run with the model's higher quality setting rather than editing the prompt endlessly.
• You can't edit afterwards. Correct - you've left design coordinates. Keep the pre-presentation master in a Badge Edit State Save snapshot; that's the branchable state.
• It's only one prompt. If you want the deterministic version of the same idea - white background, blurred shadow, edge light, no model in the loop - that's Badge Studio Composite V1, which does it with maths instead of a prompt and never drifts.