Tutorial and Prompts

All Stable Diffusion inference was done using SD WebUI Forge Neo on an RTX 3050 6gb, RTX 3060 12gb, and RTX 4060 8gb.
medium

Keyframe Images

Workflow Diagram:
large

Step 1 - Posing

Model: Prefect Illustrious XL V7
Set up the character, pose, and composition during this step.
Keyframe 1 Parameters:

[score_9, score_8_up, score_7_up, score_6_up], fur_tuft, Rarity, female anthro pony, booty shorts, walking, butt, hip dip, simple background
Steps: 24, Sampler: RES4LYF res_2m, Schedule type: Beta 57, CFG scale: 4.5, Seed: 1007204729, Size: 832x1216, Model: prefectIllustriousXL_v60, Model hash: a1e094d1ba, Module 1: sdxl.vae, RNG: CPU, Beta schedule alpha: 0.5, Beta schedule beta: 0.7, Emphasis: Original, Version: neo-2.24

Step 2 - Depth Estimation

Model: depth_midas
Run the image through a depth processor to guide the next step

Step 3 - Detailing

Model: Pony Realism 2.2
ControlNet Model: xinsir/controlnet-depth-sdxl-1.0.safetensors
Use this step for lighting and detailing. Apply the ControlNet model using Step 2 as the input. I used Strength: 0.2, Control Steps: 0-0.8.
Keyframe 1 Parameters:

[score_9, score_8_up, score_7_up, score_6_up], female anthro pony, g4, fur_tuft, [:Rarity:4], fluffy, butt , walking, dock \(anatomy\), hip dips, thighs, rear view, fashionable clothing, short shorts, simple background, rim light, haze, warm volumetric sunlight, [dark background], dust motes, film grain
Negative prompt: cute
Steps: 24, Sampler: RES4LYF res_2m, Schedule type: Beta 57, CFG scale: 2.5, Seed: 2802500537, Size: 832x1216, Model: ponyRealism_V22, Model hash: 7c97ecf786, Module 1: sdxl.vae, Denoising strength: 0.2, RNG: CPU, ControlNet 0: "Module: depth_midas, Model: controlnet-depth-sdxl-1.0, Weight: 0.2, Resize Mode: Crop and Resize, Processor Res: 512, Threshold A: 0.5, Threshold B: 0.5, Guidance Start: 0.0, Guidance End: 0.77, Pixel Perfect: False, Control Mode: Balanced, Hr Option: Both", CG Recenter: 0.35, CG Normalization: 0, Hires Module 1: Use same choices, Hires upscale: 2.5, Beta schedule alpha: 0.5, Beta schedule beta: 0.7, Emphasis: Original, Hires upscaler: 4x-UltraSharp, Hires steps: 12, Hires CFG Scale: 4.5, Version: neo-2.24

Step 4: Background

Model: WAN 2.7 Image Pro
Throw together a background; I just snapped a picture on my phone

Video Generation

Model: Kling 3.0
Input the keyframes in frames mode and link the extra reference images.

Final Steps

Used a video editor to splice everything together, add color grading, and enhance the framerate.

Syntax quick reference: **bold** *italic* ||hide text|| `code` __underline__ ~~strike~~ ^{^sup^} _~sub~

Detailed syntax guide

# Tutorial and Prompts
All Stable Diffusion inference was done using [SD WebUI Forge Neo](https://github.com/Haoming02/sd-webui-forge-classic/tree/neo) on an RTX 3050 6gb, RTX 3060 12gb, and RTX 4060 8gb.
![medium](https://dominantmares.com/sd.jpg)

## Keyframe Images
Workflow Diagram:
![large](https://dominantmares.com/Rara.jpg)

### Step 1 - Posing
Model: [Prefect Illustrious XL V7](https://tensor.art/models/967568879796651199)
Set up the character, pose, and composition during this step.
Keyframe 1 Parameters:
```
[score_9, score_8_up, score_7_up, score_6_up], fur_tuft, Rarity, female anthro pony, booty shorts, walking, butt, hip dip, simple background
Steps: 24, Sampler: RES4LYF res_2m, Schedule type: Beta 57, CFG scale: 4.5, Seed: 1007204729, Size: 832x1216, Model: prefectIllustriousXL_v60, Model hash: a1e094d1ba, Module 1: sdxl.vae, RNG: CPU, Beta schedule alpha: 0.5, Beta schedule beta: 0.7, Emphasis: Original, Version: neo-2.24
```

### Step 2 - Depth Estimation
Model: `depth_midas`
Run the image through a depth processor to guide the next step

### Step 3 - Detailing 
Model: [Pony Realism 2.2](https://tensor.art/models/780683170187288708)
ControlNet Model: `xinsir/controlnet-depth-sdxl-1.0.safetensors`
Use this step for lighting and detailing. Apply the ControlNet model using Step 2 as the input. I used `Strength: 0.2, Control Steps: 0-0.8`.
Keyframe 1 Parameters: 
```
[score_9, score_8_up, score_7_up, score_6_up], female anthro pony, g4, fur_tuft, [:Rarity:4], fluffy, butt , walking, dock \(anatomy\), hip dips, thighs, rear view, fashionable clothing, short shorts, simple background, rim light, haze, warm volumetric sunlight, [dark background], dust motes, film grain
Negative prompt: cute
Steps: 24, Sampler: RES4LYF res_2m, Schedule type: Beta 57, CFG scale: 2.5, Seed: 2802500537, Size: 832x1216, Model: ponyRealism_V22, Model hash: 7c97ecf786, Module 1: sdxl.vae, Denoising strength: 0.2, RNG: CPU, ControlNet 0: "Module: depth_midas, Model: controlnet-depth-sdxl-1.0, Weight: 0.2, Resize Mode: Crop and Resize, Processor Res: 512, Threshold A: 0.5, Threshold B: 0.5, Guidance Start: 0.0, Guidance End: 0.77, Pixel Perfect: False, Control Mode: Balanced, Hr Option: Both", CG Recenter: 0.35, CG Normalization: 0, Hires Module 1: Use same choices, Hires upscale: 2.5, Beta schedule alpha: 0.5, Beta schedule beta: 0.7, Emphasis: Original, Hires upscaler: 4x-UltraSharp, Hires steps: 12, Hires CFG Scale: 4.5, Version: neo-2.24
```

### Step 4: Background
Model: [WAN 2.7 Image Pro](https://www.atlascloud.ai/models/alibaba/wan-2.7-pro/image-edit)
Throw together a background; I just snapped a picture on my phone

## Video Generation
Model: [Kling 3.0](https://klingai.com/global/)
Input the keyframes in frames mode and link the extra reference images.

## Final Steps
Used a video editor to splice everything together, add color grading, and enhance the framerate.

not provided yet

Comments

Bad Dragon

It’s not the destination, it’s the journey.

Posted about 4 hours ago Report

Link Quote Reply