Presented by Jingwei Ma
Related #1 - Textual Inversion
Related #2 - Dreambooth
Related #3 - UniTune
Related #3 - UniTune
Comparison
Textual Inversion
Posture/Action
Text control
Unchanged
No control
Imagic�Dreambooth
UniTune
Textual Inversion
Context/Background
Text control
Unchanged
Affected by text
Imagic
Textual Inversion�Dreambooth
Identity
Text control
Unchanged
Same concept
Unitune
Imagic�Dreambooth
UniTune
Comparison
Textual Inversion
Posture/Action
Text control
Unchanged
No control
Imagic�Dreambooth
UniTune
Textual Inversion
Context/Background
Text control
Unchanged
Affected by text
Imagic
Textual Inversion�Dreambooth
Identity
Text control
Unchanged
Same concept
Unitune
Imagic�Dreambooth
UniTune
Satisfying the target prompt while preserving maximal content from image
Comparison
Textual Inversion
Posture/Action
Text control
Unchanged
No control
Imagic�Dreambooth
UniTune
Textual Inversion
Context/Background
Text control
Unchanged
Affected by text
Imagic
Textual Inversion�Dreambooth
Identity
Text control
Unchanged
Same concept
Unitune
Imagic�Dreambooth
UniTune
Method
Overview: 3 stages
Stage1: Text Embedding Optimization
Objective:
Optimize:
After Stage 1
Not exactly the same as original image
Original image
Generated
Stage 2: Model fine-tuning
Input
Before fine-tuning
After fine-tuning
Stage3: Generation
Top row = pretrained, bottom row = fine-tuned
More results
Different prompt, same image
Same prompt, different samples
More examples
Stable Diffusion Implementations