ComfyUI / Workflow template
Keep a strong character reference and build the next look around it. This ComfyUI template connects LoadImage to Seedream 5 Pro Edit and SaveImage, with an optional second reference socket for wardrobe or scene direction. Use it for original adult fashion galleries, themed lookbooks and new backgrounds without rebuilding a local sampling graph.
Import the JSON, select your credentials and fill the required inputs before running.
Start by changing one thing: the jacket, the setting or the lighting. State which face, hairstyle, pose and framing should remain. Once the edit is useful, add a second reference or request a more ambitious composition. This is a reference-guided image edit, so the best workflow begins with a deliberate source rather than a long list of settings.
How the workflow runs
Choose the source image
Describe one edit
Generate the variation
Save the keeper
NSFW / Character / Style
Explore a character’s wardrobe without rebuilding the entire scene: a fitted evening look, a themed costume or a different fabric treatment. Reference editing lets you compare the creative change against the original.
What to change first
Describe the garment, color and fabric explicitly. Keep pose and background steady for the first comparison.
Give a recurring character a complete editorial story: window-light portraits, an evening look and a late-night balcony. Use four scenes to find a direction, then expand the strongest look into a larger set.
What to change first
Reuse the original portrait for each scene. Change the wardrobe and location; keep facial details and hair consistent.
Create a recognizable male character across relaxed portraits, silk-shirt editorials and night-out looks. Keep a few signature traits so the gallery feels like one creator rather than a collection of unrelated faces.
What to change first
Keep the same reference and hair description. Change lighting or wardrobe first, then try a different framing.
Workflow tutorial
Install the SpicyAPI extension, restart ComfyUI and import the JSON download. Add your key in Settings → SpicyAPI. This graph uses Seedream 5.0 Pro Edit, with one hosted generation task per execution and no local checkpoint loader.
Load an original fictional adult-character image with a readable face and usable pose. If you want a wardrobe change, keep the background simple at first. A clean source makes it easier to compare the outfit edit without simultaneously judging a new face, pose and environment.
Fill the empty prompt with the specific edit, followed by the face, hair, pose, framing or light you want to preserve. For the first pass, choose one change. When adding a wardrobe reference, connect another LoadImage to image_urls.image_1 and explain that it supplies clothing, while reference 1 supplies the character.
The starter uses resolution 1k and aspect_ratio auto, so your first review follows the source’s composition. The endpoint also offers 2k. Choose PNG for a saved editing asset or JPEG for a smaller opaque image. enable_thinking is enabled in the starter; it is an instruction-processing option, not a local sampler setting.
Queue once and inspect the returned image. SaveImage writes to your configured ComfyUI output directory with spicyapi/seedream-edit as its prefix. Compare the requested garment or scene, the face and the pose. Return to the original source for a clean new variation; use a previous result when your intention is to refine that particular version.
Creative example. Adapt it to your own fictional adult character and source image.
Keep the fictional adult character's face, copper hairstyle, pose, framing and light direction from the source. Change only the outfit to an elegant burgundy satin evening dress with simple gold accessories. Preserve the same background and natural expression.Try this directionUse Seedream 5 Pro Edit for wardrobe and scene variations from a strong reference. Keep the original image as the anchor across a lookbook, then select your best edit before making more variants.
Each queue submits one hosted edit task. The quote reflects the actual references and selected settings; review the current price on the model page and set Max Cost Per Run accordingly. Start with one source at 1k for review, then choose 2k when the selected variation needs a larger final image. This template does not add an automatic upscale stage or silently submit a second generation.
Reduce the edit to the garment, explicitly preserve face and hair, and return to the original source. If a clothing reference is attached, state that it contributes the outfit rather than the person.
Keep the preservation instruction concrete: same background, pose, camera angle and light direction. Avoid introducing a new location in the same first-pass prompt.
Inspect the IMAGE link into SaveImage and the configured output directory. A preview or completed hosted task does not by itself prove that the local saver ran. Use the task ID to recover a result if the local wait was interrupted.
LoadImage
Describe one edit
Qwen reference edit
SaveImage
ComfyUI
Outfit variants, scene changes and character frames
An edited image saved by SaveImage
Open workflowReference + scenes
One item per scene
Quote + Qwen edit
Download each image
n8n Cloud + self-hosted
Recurring character sets and NSFW creator galleries
4, 8 or 12 separate image results
Open workflowDownload the template or try the same model in the Playground before connecting the full workflow.