Small artifacts in HDR Image about hdr-imaging

[ssdutHB]

yes, in my post, the result (tonemapped) have artifacts also, it is still about tonemapping issue. it is not a easy task when converting HDR to SDR. You may need to implement other tonemap algorithms for this.

How can you know that the hdr itself does not have them?

Because of the HDR radiance map didn't have those unnormal noises, I still tend to believe the issue is due to tonemap algorithm.

When I directly passed the hdr array to tone mapping functions implemented in cv2, the result looked good. But if I saved the array mentioned above as .hdr file with the function save_hdr, and read the image and applied the tone mapping again, there were artifacts. So, It is clear that the save_hdr function leads to the artifacts. And substituting the save_hdr function with cv2.imwrite("res.hdr", hdr) sovle the problem.

[SSARCandy]

it is due to tonemapper, not hdr itself. you can refer to the radiance map in my ipython notebook https://github.com/SSARCandy/HDR-imaging/blob/master/HDR-playground.ipynb
It doesn't have those artifacts.

[tobiasmueller14]

OK, can you recommend a suitable tonemapper or parameters? Only for the meomrial image with lambda = 50.
Thank you

[tobiasmueller14]

https://ssarcandy.tw/2017/04/16/High-Dynamic-Range-Imaging/ there are the same artifacts. How can you know that the hdr itself does not have them?

[SSARCandy]

yes, in my post, the result (tonemapped) have artifacts also, it is still about tonemapping issue. it is not a easy task when converting HDR to SDR. You may need to implement other tonemap algorithms for this.

How can you know that the hdr itself does not have them?

Because of the HDR radiance map didn't have those unnormal noises, I still tend to believe the issue is due to tonemap algorithm.

[SSARCandy]

thank you for your research, I'll dig into it