The inference results are of poor quality.

[wcxfbq]

When I run the inference command "python sample_cond_ldm.py --cfg ./configs/BSDS_sample.yaml", the generated results are very blurry, and all the weights used are pre-trained. Why is this happening?

[GuHuangAI]

The results look like they were generated by the random-initialized weight.
Please make sure that u loaded the correct path of the pre-trained weights.

[wcxfbq]

Thank you very much for your response. I can confirm that the weights for both stages have been downloaded correctly, as when I run the command `python demo.py --input_dir $your input dir$ --pre_weight $the downloaded weight path$ --out_dir $the path saves your results$ --bs 8`, the generated images are no longer blurry.I find it quite strange. Could it be that there is an issue with the code in `sample_cond_ldm.py`?

…

---Original--- From: ***@***.***> Date: Sat, Jan 25, 2025 18:30 PM To: ***@***.***>; Cc: ***@***.******@***.***>; Subject: Re: [GuHuangAI/DiffusionEdge] The inference results are of poorquality. (Issue #41) The results look like they were generated by the random-initialized weight. Please make sure that u loaded the correct path of the pre-trained weights. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***>