Wan-AI/Wan2.1-T2V-14B
Text-to-Video
ā¢
Updated
ā¢
108k
ā¢
ā¢
1.15k
dylan
starkx
AI & ML interests
llms ludwig/deepspeed/ray
Recent Activity
liked
a model
4 days ago
Wan-AI/Wan2.1-T2V-14B
liked
a Space
2 months ago
hexgrad/Kokoro-TTS
liked
a model
2 months ago
deepseek-ai/DeepSeek-R1-Distill-Qwen-32B
Organizations
starkx's activity
liked
a
model
4 days ago
liked
a
Space
2 months ago
liked
5
models
2 months ago
deepseek-ai/DeepSeek-R1-Distill-Qwen-32B
Text Generation
ā¢
Updated
ā¢
2.28M
ā¢
ā¢
1.31k
hexgrad/Kokoro-82M
Text-to-Speech
ā¢
Updated
ā¢
1.87M
ā¢
3.92k
deepseek-ai/DeepSeek-R1
Text Generation
ā¢
Updated
ā¢
1.4M
ā¢
ā¢
11.8k
THUDM/glm-4-9b
Text Generation
ā¢
Updated
ā¢
9.54k
ā¢
128
jinaai/ReaderLM-v2
Text Generation
ā¢
Updated
ā¢
54.7k
ā¢
ā¢
597
reacted to
sanaka87's
post with š„
3 months ago
Post
1758
š Excited to Share Our Latest Work: 3DIS & 3DIS-FLUX for Multi-Instance Layout-to-Image Generation! ā¤ļøā¤ļøā¤ļø
šØ Daily Paper: 3DIS-FLUX: simple and efficient multi-instance generation with DiT rendering (2501.05131)
š Code is now open source!
š Project Website: https://limuloo.github.io/3DIS/
š GitHub Repository: https://github.com/limuloo/3DIS
š 3DIS Paper: https://arxiv.org/abs/2410.12669
š 3DIS-FLUX Tech Report: https://arxiv.org/abs/2501.05131
š„ Why 3DIS & 3DIS-FLUX?
Current SOTA multi-instance generation methods are typically adapter-based, requiring additional control modules trained on pre-trained models for layout and instance attribute control. However, with the emergence of more powerful models like FLUX and SD3.5, these methods demand constant retraining and extensive resources.
āØ Our Solution: 3DIS
We introduce a decoupled approach that only requires training a low-resolution Layout-to-Depth model to convert layouts into coarse-grained scene depth maps. Leveraging community and company pre-trained models like ControlNet + SAM2, we enable training-free controllable image generation on high-resolution models such as SDXL and FLUX.
š Benefits of Our Decoupled Multi-Instance Generation:
1. Enhanced Control: By constructing scenes using depth maps in the first stage, the model focuses on coarse-grained scene layout, improving control over instance placement.
2. Flexibility & Preservation: The second stage employs training-free rendering methods, allowing seamless integration with various models (e.g., fine-tuned weights, LoRA) while maintaining the generative capabilities of pre-trained models.
Join us in advancing Layout-to-Image Generation! Follow and star our repository to stay updated! ā
šØ Daily Paper: 3DIS-FLUX: simple and efficient multi-instance generation with DiT rendering (2501.05131)
š Code is now open source!
š Project Website: https://limuloo.github.io/3DIS/
š GitHub Repository: https://github.com/limuloo/3DIS
š 3DIS Paper: https://arxiv.org/abs/2410.12669
š 3DIS-FLUX Tech Report: https://arxiv.org/abs/2501.05131
š„ Why 3DIS & 3DIS-FLUX?
Current SOTA multi-instance generation methods are typically adapter-based, requiring additional control modules trained on pre-trained models for layout and instance attribute control. However, with the emergence of more powerful models like FLUX and SD3.5, these methods demand constant retraining and extensive resources.
āØ Our Solution: 3DIS
We introduce a decoupled approach that only requires training a low-resolution Layout-to-Depth model to convert layouts into coarse-grained scene depth maps. Leveraging community and company pre-trained models like ControlNet + SAM2, we enable training-free controllable image generation on high-resolution models such as SDXL and FLUX.
š Benefits of Our Decoupled Multi-Instance Generation:
1. Enhanced Control: By constructing scenes using depth maps in the first stage, the model focuses on coarse-grained scene layout, improving control over instance placement.
2. Flexibility & Preservation: The second stage employs training-free rendering methods, allowing seamless integration with various models (e.g., fine-tuned weights, LoRA) while maintaining the generative capabilities of pre-trained models.
Join us in advancing Layout-to-Image Generation! Follow and star our repository to stay updated! ā
liked
a
model
3 months ago
commented on
Train 400x faster Static Embedding Models with Sentence Transformers
3 months ago
Nvidia: You need to buy GPUs and machines.
Developer: No, we can just tweak the algorithms.
Elon Musk: Look at me buying GPUs, you poor folks.
upvoted
a
collection
3 months ago
upvoted
a
paper
3 months ago
updated
a
collection
3 months ago
liked
a
model
3 months ago
updated
a
collection
3 months ago
liked
2
models
3 months ago
liked
a
Space
3 months ago
liked
2
models
3 months ago