ResourcesVideo 03
The workflows, prompts and model links from the video.
Tested on ComfyUI 0.39.2. Use your own pictures, or the example pictures from the video (11.2 MB).
01
A prompt in, a picture out.
A vertical 35mm film portrait of a young woman with short black hair in a cream knit sweater on a ferry deck, wind in her hair, grey-blue sea behind her. Soft overcast light, muted film colours, natural skin.
02
Starts from a photo. Denoise sets how much of it changes.
A wide candid street photograph of an old man with a white beard, round glasses and a brown flat cap waiting at a busy city crossroads, holding a bunch of yellow sunflowers. Behind him a green double-decker bus, a corner cafe with a green striped awning and a sign that says CAFE ROMA, a yellow taxi and people on the zebra crossing. Warm late-afternoon sun, 35mm film look.
crossroads.png (PNG, 2.0 MB)
03
Changes one thing in a photo from a written instruction.
Make the bus green. Keep everything else the same.
crossroads.png (PNG, 2.0 MB)
04
Puts the same person in a new scene, or turns her into a character sheet.
Turn this photo into a character reference sheet showing front, three-quarter, side and back views on a simple background, waist up. Keep the same character, hair and outfit. No text or labels.
Put this woman in a cozy cafe, sitting by the window with a cup of coffee. Keep the same person, hair, and outfit.
her.png (PNG, 1.7 MB)
05
Makes the picture four times bigger and saves it.
Turn this photo into a character reference sheet showing front, three-quarter, side and back views on a simple background, waist up. Keep the same character, hair and outfit. No text or labels.
her.png (PNG, 1.7 MB)
06
Four pictures in one edit: her, the dress, the bag and the shoes.
The woman from image 1 wearing the dress from image 2, carrying the bag from image 3 and wearing the shoes from image 4. Keep her face, curly auburn hair, pose and the street the same.
07
Adds a clay style LoRA to the edit.
Transform the image into a clay sculpture style.
her.png (PNG, 1.7 MB)
08
Changes only the part you mask.
Replace the area marked white in image 2 with a black leather biker jacket. Keep everything else in image 1 exactly the same.
her.png (PNG, 1.7 MB)Right-click Load Image, Open in MaskEditor, paint over the jacket with the mask tool and save.
Download JSON for Inpainting
09
Removes whatever you paint over in green.
Remove the object marked in green and fill the area naturally using the surrounding scene. Keep everything else the same.
her.png (PNG, 1.7 MB)In MaskEditor, paint over her with the brush in bright green at opacity 1, then save.
Download JSON for Remove objects
10
Extends a picture to the left and right.
Extend both sides with more ferry deck, railing, sea, and a distant coastline. Keep the woman and the center of the image the same.
11
Takes the pose from one photo for a new picture.
A photograph of an astronaut in a white spacesuit leaping across the grey dusty surface of the moon, the blue Earth hanging in the black sky.
dancer.png (PNG, 1.2 MB)
12
Keeps a room's layout and changes its style.
A Scandinavian living room with warm evening light.
living-room.png (PNG, 1.2 MB)Put each file in its folder, then press R in ComfyUI.
qwen_image_2.1_int8_convrot.safetensors Qwen Image 2.1, the image model.models/diffusion_models/Size7.26 GBNeeded forEvery workflowqwen3vl_8b_int8_convrot.safetensors The text encoder.models/text_encoders/Size9.35 GBNeeded forEvery workflowqwen_image_2.1_vae_bf16.safetensors The VAE.models/vae/Size676 MBNeeded forEvery workflow4x-UltraSharp.pthThe upscaler.models/upscale_models/Size67 MBNeeded for05 UpscaleQwen-Image-2.1-Clay-Sculpture-Style-3000.safetensorsThe clay style LoRA.models/loras/Size84 MBNeeded for07 LoRAqwen_image_2.1_fun_controlnet_union_int8_convrot.safetensors The ControlNet model.models/model_patches/Size3.78 GBNeeded for11 and 12 ControlNetWorkflows 11 and 12 also need the ControlNet Aux custom nodes: in ComfyUI, click Extensions, search controlnet aux, install and restart.