cswry OSEDiff issues - Githubissues

cswry / OSEDiff

[NeurlPS2024] One-Step Effective Diffusion Network for Real-World Image Super-Resolution

Apache License 2.0

215 stars 12 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

We noticed that some images become more colorful and vibrant after being processed by OSEDiff. Is this a controllable option?

#49 ch3nboyu opened 6 days ago
0
Unable to Train Model on RTX 4090 Due to Out of Memory Issue

#48 ZHT150798 opened 1 week ago
1
why ram acts on label not input？

#47 Alixing opened 2 weeks ago
0
About GPU Version

#46 nitec427 opened 2 weeks ago
0
我尝试从checkpoint继续训练，但LPIPS_loss还是在1以上看起来和从头寻没区别，这正常吗？

#45 YuChen-Liang closed 2 weeks ago
0
VSD loss NaN

#44 zzzzzero closed 3 weeks ago
4
Repo id must be in the form 'repo_name' or 'namespace/repo_name': 'weights/SD21_Base/v2-1_512-ema-pruned.ckpt'

#43 2590609710 opened 4 weeks ago
2
Question about VSD loss

#42 Renzhihan closed 4 weeks ago
1
result resolution

#41 milely closed 1 month ago
0
Bug in new update? Windows OS. previous commit work OK

#40 zelenooki87 closed 1 month ago
2
What are the pre-train weights for the diffusion network and the 2 regularizers respectively?

#39 YaqiWangCV closed 1 month ago
1
what does the VSD loss in image domain means?

#38 heart-du opened 1 month ago
0
Issue with ram/models/utils.py functiom init_tokenizer()

#37 mashoutsider opened 2 months ago
1
about the GPU memory used in the training process

#36 beyondbatman-master closed 1 month ago
5
Update on the repository's license

#35 Drajan closed 2 months ago
2
only PSNR and SSIM

#34 zzctmd opened 2 months ago
1
Blur issue happened on the flatten region

#33 tommy830725 closed 1 month ago
1
The results are "unfairly" much more natural and detailed without the tiling function

#32 zelenooki87 closed 2 months ago
4
About the stability of finetuning vae with LoRA

#31 icandle closed 1 month ago
3
Reproduce the training process for multiple losses encounters "RuntimeError: Expected to mark a variable ready only once"

#30 Yangkai-Wei closed 2 months ago
0
AttributeError: 'UNet2DConditionModel' object has no attribute 'add_adapter'.

#29 1343464520 closed 2 months ago
4
OSEDiff video upscaling script

#28 zelenooki87 closed 1 month ago
4
Fix osediff_path and BertTokenizer path

#27 zhtjtcz opened 3 months ago
0
请问支持diffusers包升级到最新吗？

#26 1343464520 opened 3 months ago
1
Support for FLUX?

#25 AIisCool opened 3 months ago
3
Is there any difference between your LoRA and img2img turbo?

#24 MontaEllis closed 1 month ago
1
关于网络使用的问题

#23 lizhaohu closed 1 month ago
1
Calculate the metrics problem

#22 sacred178953 closed 1 month ago
1
Some confusion about VSD loss implementation

#21 zzzzzuber opened 4 months ago
14
windows error：expected str, bytes or os.PathLike object, not NoneType

#20 ZhenyanSun closed 4 months ago
1
Is Lora training necessary?

#19 Feynman1999 opened 4 months ago
1
When will you release training codes?

#18 lyf1212 closed 4 months ago
1
Question about MANIQA Metric

#17 wangjg33 closed 4 months ago
1
questions about the output.

#16 TT-T1 closed 4 months ago
1
How these indicators such as PSNR, SSIM and FID are calculated?

#15 sacred178953 closed 4 months ago
1
when trying to adapt to sdxl I am unable to properly add time_ids to UNet2DConditionModel

#14 saiFarmer closed 1 month ago
1
Whether release training code or not?

#13 zzzzzuber closed 4 months ago
5
Question: SD3 support

#12 AIisCool opened 4 months ago
0
About the model folder

#11 wangjg33 closed 4 months ago
1
About the implementation of the method

#10 ChenhLiwnl opened 4 months ago
1
Output results not as good as SeeSR?

#9 AIisCool closed 4 months ago
4
fp16inference + tile option?

#8 zelenooki87 closed 4 months ago
7
huggingface_hub.errors.HFValidationError: Repo id must be in the form 'repo_name' or 'namespace/repo_name': '/home/notebook/data/group/LowLevelLLM/LLM/bert-base-uncased'. Use `repo_type` argument if needed.

#7 zelenooki87 opened 4 months ago
9
SDXL-Turbo?

#6 AIisCool closed 4 months ago
2
Question: Is the time step directly fixed to T in the input of diffusion (the largest time step)

#5 Feynman1999 closed 4 months ago
5
Question: Able to make this temporally stable for video input?

#4 AIisCool closed 4 months ago
1
Excited but hope issues from SeeSR are addressed

#3 AIisCool closed 4 months ago
1
May I ask when the author will upload the complete code

#2 paomian001 closed 4 months ago
1
code and weights

#1 flynnamy closed 4 months ago
1