No need for models, use Stable Diffusion to generate AI clothing model scene images at low cost

This tutorial mainly uses Stable Diffusion This tool is used to operate. How can we use Stable Diffusion to assist us in completing the model effect display? This tutorial is suitable for e-commerce design scenes, photography scenes and other actual combat using character design. The whole process is practical and requires everyone to absorb it slowly. After learning it, you can easily control the model's outfit change. Let's go!

Vincent Figure

Libulibu is generated online. First use Wenshengtu, open the website, follow the rhythm, and start making

https://www.liblib.art/sd

This tutorial is written in order from top to bottom.

① Select the large model, ② Enter keywords, ③ Set the drawing parameters, ④ Set ControlNet

1. Choose a good large model

Large Model: Maiju Realism

Large models can be searched here, just click to add to the model library

https://www.liblib.art

No need for models, use Stable Diffusion to generate AI clothing model scene images at low cost

2. Enter keywords

SD keywords are divided into positive and negative

Positive keywords:

Half of a beautiful woman,Outdoor,background with sky,White halter dress,long black hair,The wind blows the hair,hair flowing,best quality,masterpiece,eyes closed,breeze blowing hair,background blurred,5 fingers,Natural light, blue sky, sky with no other object

Half-length portrait of a beautiful woman, outdoors, sky background, white suspender dress, long black hair, wind blowing her hair, flowing hair, best quality, masterpiece, closed eyes, breeze blowing her hair, blurred background, 5 fingers, natural light, blue sky, sky with no other objects

Negative keywords:

ng_deepnegative_v1_75t,(badhandv4:1.2),EasyNegative,(worst quality:2),easynegative,drawing,painting,crayon,sketch,graphite,impressionist,noisy,blurry,soft,deformed,ugly,(deformed iris, deformed pupils, semi- realistic, cgi, 3d, render, sketch, cartoon, drawing, anime:1.4),text,close up,cropped,out of frame,worst quality,low quality,jpeg artifacts,ugly,duplicate,morbid,mutilated,extra fingers, mutated hands,poorly drawn hands,poorly drawn face,mutation,deformed,blurry,dehydrated,bad anatomy,bad proportions,extra limbs,cloned face,disfigured,gross proportions,malformed limbs,missing arms,missing legs,extra arms,extra legs,fused fingers,too many fingers,long neck,((nsfw)),sketches,tattoo,(beard:1.3),(EasyNegative:1.3),badhandv4,(worst quality:2),( low quality:2),(normal quality:2),lowers,normal quality,facing away,looking away,text,error,extra digit,fewer digits,cropped,jpeg artifacts,signature,watermark,username,blurry,skin spots, acnes,skin blemishes,bad anatomy,fat,bad feet,cropped,poorly drawn hands,poorly drawn face,mutation,deformed,tilted head.bad anatomy.bad hands,extra fingers,fewer digits,extra limbs.extra arms,extra legs ,malformed limbs.fused fingers,too many fingers,long neck,cross-eyed,mutated hands,bad body,bad proportions,gross proportions,text,error,missing fingers,missing arms,missing legs,extra digit,extra arms,extra leg,extra foot,missing fingers,(Worst quality,low quality,lowres:1.2) ,error,cropped,jpeg artifacts,out of frame,watermark,signature

ng_deepnegative_v1_75t, (badhandv4:1.2), EasyNegative, (worst quality:2), EasyNegative, drawing, painting, crayon, sketch, graphite, impressionist, noise, blur, soft, distorted, ugly, (distorted iris, distorted pupil, semi-realistic, cgi, 3d, render, sketch, cartoon, painting, animation:1.4), text, close up, cropped, out of frame, worst quality, low quality, jpeg artifacts, ugly, duplicate, sick, mutilated, extra fingers, mutant hand, badly drawn hand, badly drawn face, mutation, deformed, blurry, dehydrated, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, thick proportions, deformed limbs, missing arm, missing leg, extra arm, extra leg, fused fingers, too many fingers, long neck, ((ns fw), sketch, tattoo, (beard: 1.3), (EasyNegative: 1.3), badhand 4, (worst quality: 2), (low) quality: 2), (normal quality: 2), low, normal quality, facing away, looking away, text, error, extra digits, fewer digits, cropped, jpeg artifacts, signature, watermark, username, blurry, skin spots, acne, skin blemishes, bad anatomy, fat, bad feet, cropped, badly drawn hands, badly drawn face, mutation, deformed, tilted head. bad anatomy. bad hands, too many fingers, too few fingers, too many limbs. extra arms, extra legs, misshapen limbs. fused fingers, too many fingers, long neck, crossed eyes, mutant hands, bad body, bad proportions, thick proportions, text, errors, missing fingers, missing arms, missing legs, extra fingers, extra arms, extra legs, extra feet, missing fingers, (worst quality, low quality, low resolution: 1.2), errors, cropped, jpeg artifacts, out of frame, watermark, signature

3. Set the drawing parameters

Sampling method: DPM++ 2M Karras

Iterations: 30

Check High Resolution Repair

Amplification algorithm: 4x-UltraSharp

Redraw sampling steps: 20

Redraw amplitude: 0.4

Proportional scaling: 2

Number of pictures: 3 (you can add more if the effect test is satisfactory)

Prompt wordGuidance factor: 7

No need for models, use Stable Diffusion to generate AI clothing model scene images at low cost

4. Set up ControlNet

Upload the reference image here and check Enable

ControlNet chooses OpenPose to fix character poses

Preprocessor selection: dw_openpose_full

Model selection: control_v11p_sd15_openpose

Control weight: 1

Starting steps: 0

Steps completed: 1

No need for models, use Stable Diffusion to generate AI clothing model scene images at low cost

Upload the reference image here and check Enable

ControlNet selects the Depth map to control the finger

Preprocessor selection: depth_hand_refiner

Model selection:

control_sd15_inpaint_depth_hand_fp16

Control weight: 1

Starting steps: 0

Steps completed: 1

No need for models, use Stable Diffusion to generate AI clothing model scene images at low cost

Once everything is done, you can start generating images. View the generated images in the gallery.

No need for models, use Stable Diffusion to generate AI clothing model scene images at low cost

I looked through a lot of pictures here, trying to make the fingers look good in one go, but the result did not meet my requirements, so I decided to choose one with similar effect first, and then retouch the hands in the raw picture.

No need for models, use Stable Diffusion to generate AI clothing model scene images at low cost

Figure 1

Now I start using the image generation function, the main purpose is to repair the hand

① Select the large model, ② Enter keywords, ③ Select the LORA model, ④ Upload the image, ⑤ Set the output parameters, ⑥ Set ControlNet

1. Choose a good large model

Large Model: Maiju Realism

Large models can be searched here, just click to add to the model library

https://www.liblib.art

No need for models, use Stable Diffusion to generate AI clothing model scene images at low cost

2. Enter keywords

Positive keywords:

Hands,good-looking hands,perfect hands,(five fingers:1.1),(Nails intact:1.1),finger repair,Finger optimization,repair Hand optimization,complete repair

Hands, beautiful hands, perfect hands, (five fingers: 1.1), (intact nails: 1.1), finger repair, finger optimization, repair hand optimization, complete repair

Negative keywords:

Deformed,deformed hands,four fingers,six fingers

Deformed, deformed hand, four fingers, six fingers

3. Choose a good LORA model

Click Add

Model 1: Hand optimization - hand repair is delicate and realistic, with the weight set to 0.8

LORA models can be searched here, just click to add to the model library

https://www.liblib.art

No need for models, use Stable Diffusion to generate AI clothing model scene images at low cost

4. Upload pictures

Upload the image saved in the Wenshengtu to the Shengtu → Partial Redraw, and use the brush to blacken the hand area

Mask Mode: Redraw the mask content

No need for models, use Stable Diffusion to generate AI clothing model scene images at low cost

5. Set the drawing parameters

Sampling method: DPM++ 2M Karras

Iterations: 25

Resize: 1.3 times

Number of images: 3

Prompt word guidance coefficient: 7

Redraw step: 1

No need for models, use Stable Diffusion to generate AI clothing model scene images at low cost

6. Set up ControlNet

Upload the pictures saved in Wenshengtu here and check Enable

ControlNet chooses OpenPose to fix character poses

Preprocessor selection: dw_openpose_full

Model selection: control_v11p_sd15_openpose

Control weight: 1

Starting steps: 0

Steps completed: 1

No need for models, use Stable Diffusion to generate AI clothing model scene images at low cost

Upload the pictures saved in Wenshengtu here and check Enable

ControlNet selects the Depth map to control the finger

Preprocessor selection: depth_hand_refiner

Model selection:

control_sd15_inpaint_depth_hand_fp16

Control weight: 0.65

Starting steps: 0

Steps completed: 1

No need for models, use Stable Diffusion to generate AI clothing model scene images at low cost

Once everything is done, you can start generating images. View the generated images in the gallery.

No need for models, use Stable Diffusion to generate AI clothing model scene images at low cost

The process of hand-editing is to continuously generate pictures that feel good, and then run the pictures again and again according to the process of picture generation and partial redrawing until the effect is satisfactory. To be honest, I am not very satisfied with the final picture, but I don’t want to draw cards anymore, so I decided to use PS to revise it.

No need for models, use Stable Diffusion to generate AI clothing model scene images at low cost

PS Post

① Make the characters look better; ② Add the product and handle the light and shadow; ③ Add the copy

Although the hands still don't look very good, the production process is like this. It depends on your patience. Use SD to slowly brush the pictures and PS to slowly edit them.

No need for models, use Stable Diffusion to generate AI clothing model scene images at low cost

Okay, this sharing ends here.

statement:The content is collected from various media platforms such as public websites. If the included content infringes on your rights, please contact us by email and we will deal with it as soon as possible.
TutorialEncyclopedia

AI generates free photos. Use Coze to make a complete photo album at home in 10 minutes

2024-7-13 12:24:20

Encyclopedia

ComfyUI zero-based introductory tutorial, how to build a controlNet workflow

2024-7-14 10:27:37

Search