How to calibrate the alpha/beta params and how to match them with this project?

[renwoxing000]

Hi Jamy:

I try to run this project with my raw images, the output image is blur and has some artifacts; Is the problem of params alpha/beta?
How can i match them with this project? How to calibrate the alpha/beta params?

Looking forward to your reply.

[Jamy-L]

Hi,

Artefacts can have many different causes, and I guess you are mentionning the bright colors next sharp edges. It seems to be inherent to the method, and is boosted by the post-processing. We found that applying this method could remove them efficiently.

For the blury part, you could use a stronger unsharp mask as post precessing. Blurry spots can also indicate that too many frames where rejected, which could be either due to a bad frame alignment or a wrong decision of the robustness module. If it comes from the rejection (which you can check by running the algorithm without the robustness mask), you can try to get alpha and beta using noise profiling methods like this.

My guess is that some tricky spots can get blured because the implementation is not perfect, the alignment module and the robustness mask can probably be tweaked to have better results.

Best,

Jamy

[renwoxing000]

Hi Jamy:

Thanks for your reply.
I found jagged edges、colored stripes and some discontinuous edge. What causes these artifacts?

Best,

ziro.

[y2k0999]

Hi,

Artefacts can have many different causes, and I guess you are mentionning the bright colors next sharp edges. It seems to be inherent to the method, and is boosted by the post-processing. We found that applying this method could remove them efficiently.

For the blury part, you could use a stronger unsharp mask as post precessing. Blurry spots can also indicate that too many frames where rejected, which could be either due to a bad frame alignment or a wrong decision of the robustness module. If it comes from the rejection (which you can check by running the algorithm without the robustness mask), you can try to get alpha and beta using noise profiling methods like this.

My guess is that some tricky spots can get blured because the implementation is not perfect, the alignment module and the robustness mask can probably be tweaked to have better results.

Best,

Jamy

how can I know how many frames were rejected?

[Jamy-L]

Hi, the frame rejection is enforced by the robustness mask of each frame, which is basically a fusion weight map between 0 and 1 (0 means that the frame is locally totally rejected, and 1 totally merged). You can either visualise each frame map independently (I dont think the code returns them, you might need to adjust the code), or inspect the accumulated robustness mask that can be returned using the right settings. This mask is the sum of all robustness maps, so 0 means only the reference frame is locally used, and x means that roughly x frames were locally merged.