How to denoise images with specific noise?

[D1z2y]

First, thanks for the contributions!

I was wondering how to denoise images with specific noise, like gaussian/poisson/speckle ?

In #55 About denoising, I used the code :
python main.py --ni --config celeba_hq.yml --path_y celeba_hq --eta 0.85 --deg "denoising" --deg_scale 1 --sigma_y 0.2 -i celeba_sr_ap_16_n_02 --add_noise for denosing.

Using CelebA testset and model:
How to denoise a image with one certain noise or mixed noises? For example, if I manually add poisson noise to a image in the given testset, how can I run the denoising task?

Am I supposed to use the "DIY Mode" and adjust any settings or parameters?
Could you please offer a solution to that problem? Thank you for your time and help!

[Yinhuai Wang]

Hi, the denoising task relies on the manual setting of noise levels, you can add Poisson noise and try different --sigma_y 0.2.

[D1z2y]

Hi, the denoising task relies on the manual setting of noise levels, you can add Poisson noise and try different --sigma_y 0.2.

Thanks for replying!
In code
python main.py --ni --config celeba_hq.yml --path_y celeba_hq --eta 0.85 --deg "denoising" --deg_scale 1 --sigma_y 0.2 -i celeba_sr_ap_16_n_02 --add_noise
I tried to add -n poisson before --add_noise but it seems to add gaussian noise.

The upper one is poisson, the lower one is gaussian. But they seem the same to each other.