Danne's crop_rec_4k experiments for EOS M

Started by Danne, December 03, 2018, 06:10:17 PM

Previous topic - Next topic

0 Members and 4 Guests are viewing this topic.

cedricp

I made some tweaks in the dualiso code, it fixes my filckering clips without baking and seems to not alter image (expo slope values remain very close to original code without large jumps between images). If you can try it and tell me if it's OK for you too...

iaburn

Quote from: cedricp on March 17, 2023, 10:34:29 PM
I made some tweaks in the dualiso code, it fixes my filckering clips without baking and seems to not alter image (expo slope values remain very close to original code without large jumps between images). If you can try it and tell me if it's OK for you too...

I have a couple of "flickering" clips from previous posts that seems to flicker still, maybe you can give it a try also:
https://mega.nz/file/JptAECbS#BXImCPVcmdXe6UXIXspeyZzNhy5kJWD1-aK39UyuJvU
https://mega.nz/file/pkkDkIDZ#e9aHgJtdIaYyiESYEbAXgoNifEO9QxIg85tmOgjXL0Y

cedricp

I spent some hours to try to find a better solution, but I failed :(
I removed some omp directive that caused wrong results and added more caching stuff, so that should be slightly faster to export.
The "baked" expo matching remains the best fix for that so far.

iaburn

Quote from: cedricp on March 20, 2023, 11:06:02 PM
I spent some hours to try to find a better solution, but I failed :(
I removed some omp directive that caused wrong results and added more caching stuff, so that should be slightly faster to export.
The "baked" expo matching remains the best fix for that so far.

The bake button is a good compromise for now. Can you make a pull request to my fork with your changes? I'll take a look at the omp directives that you had to remove, maybe I we can fix them  :)

cedricp

I found the ML's dual iso pdf doc on the internet that helped me to better understand the code and I finally got better results for the exposure ISO matching function. It should also be slightly faster.
For those interested the doc is here
I pushed the code on my gitub if you wanna try. It's not perfect, but it solves almost all the problematic shots.

ML700D

Quote from: cedricp on March 27, 2023, 10:04:46 PM
I found the ML's dual iso pdf doc on the internet that helped me to better understand the code and I finally got better results for the exposure ISO matching function. It should also be slightly faster.
For those interested the doc is here
I pushed the code on my gitub if you wanna try. It's not perfect, but it solves almost all the problematic shots.
nice finding.. I don't understand coding
I hope ML dual iso getting better next..

btw, is it for eos M user only?
EOS 700D

Danne

Quote from: cedricp on March 27, 2023, 10:04:46 PM
I found the ML's dual iso pdf doc on the internet that helped me to better understand the code and I finally got better results for the exposure ISO matching function. It should also be slightly faster.
For those interested the doc is here
I pushed the code on my gitub if you wanna try. It's not perfect, but it solves almost all the problematic shots.
Nice work.
Are these the changes? Anything else modified?
https://github.com/cedricp/MLV-App/commit/0028db617a9efb9c5a5933c487e05d8c5547674c

Possible to fix original source code in here?
https://bitbucket.org/Dannephoto/magic-lantern_jip_hop_git/src/master/modules/dual_iso/dual_iso.c

cedricp

Quote from: ML700D on March 28, 2023, 02:39:04 AM
nice finding.. I don't understand coding
I hope ML dual iso getting better next..

btw, is it for eos M user only?
Thanks, no, it's a fix for dual iso blending, so for all cameras supporting dual ISO.

Quote from: Danne on March 28, 2023, 08:06:49 AM
Nice work.
Are these the changes? Anything else modified?
https://github.com/cedricp/MLV-App/commit/0028db617a9efb9c5a5933c487e05d8c5547674c

Possible to fix original source code in here?
https://bitbucket.org/Dannephoto/magic-lantern_jip_hop_git/src/master/modules/dual_iso/dual_iso.c
Yes changes are there, and the code in the ML firmware is OK so far. The changes are for postprocess only.
It seems that fullres reconstruction works better too.

Danne

Quote from: cedricp on March 28, 2023, 09:24:08 AM
Yes changes are there, and the code in the ML firmware is OK so far. The changes are for postprocess only.
Yes of course, my mistake. I mean put in the code in here:
https://bitbucket.org/Dannephoto/magic-lantern_jip_hop_git/src/master/modules/dual_iso/cr2hdr.c

cedricp

Interestingly, cr2hdr has a different (maybe more robust) ISO matching function, trying it right now :)
Edit : That method is the best I tested ! No flickering at all and very good dynamic range staging.

Danne

Good. The code is the latest I believe, from a1ex. Would be great if this code is reviewed in whole and implemented in Mlv app :).

cedricp

I did a quick review. I've adapted and updated my MLVapp repo with the updates from cr2hdr command line utility. The only missing feature is noise computation. Indeed a CR2 (digital still) file seems to contain some special rows/columns of black bars to measure noise, it does not exist in a video clip because of the crop I guess.
I'll do a more accurate analysis of the a1ex code to check if all is up2date.

Danne


iaburn

cedricp I just merged your improvements on my fork to test it. Do you think that we can get rid of the "bake" button after your last commit?

cedricp

Not necessarily, cause you can get faster export if enabled though.

iaburn

Maybe a bit off topic, but what is the reason for having only 1x (crop) and 3x (standard) modes available?
What's the technical limitation for something like a 2.5K 2x mode?

masc

Quote from: iaburn on April 04, 2023, 09:47:12 AM
What's the technical limitation for something like a 2.5K 2x mode?
The bayer matrix on the sensor.
5D3.113 | EOSM.202

iaburn

Quote from: masc on April 04, 2023, 02:00:30 PM
The bayer matrix on the sensor.

Wouldn't it be possible to take every other 2 rows?
With 3x3 we read the red-green row, jump 2 rows and read the green-blue row.

Cannot we read the first 2 rows (red-green and green-blue), skip 2 rows, and read again 2 rows, so we get half of the sensor?

ML700D

Quote from: iaburn on April 04, 2023, 03:01:36 PM
Wouldn't it be possible to take every other 2 rows?
With 3x3 we read the red-green row, jump 2 rows and read the green-blue row.

Cannot we read the first 2 rows (red-green and green-blue), skip 2 rows, and read again 2 rows, so we get half of the sensor?
I'm just curious if 2x2 or 1x2 binning mode is possible..what would it be like?
EOS 700D

iaburn

Quote from: ML700D on April 04, 2023, 04:10:01 PM
I'm just curious if 2x2 or 1x2 binning mode is possible..what would it be like?

That would be like a 2.5K mode with no extra crop, and I guess aliasing and artifacts at a similar, slightly lower, level than current 1080HD mode

Edit: I've been reading some old posts and it seems like ML is using the reading modes already available in Canon's firmware. So unless there is some magic register combination to enable this, I guess this is sadly not possible u_u


crossroads

hey danne can you please fix the crop mood mess and release a new build, i ve been using your builds for a long time and i trust your ability to fix anything

K-Rim

Hey guys I'm currently on the newest Build and I have issues with my HDMI output.

I tried a SmallHD and the Atomos Ninja V Monitor, but both get no signal from the Canon EOS M.

The screen of the EOS M goes black as i put in the HDMI cable into the canon, but the screens get no signal.

I tried all ML modes and no changes.

Any guess why it doesn't work?

Thank you guys.

theBilalFakhouri

Quote from: K-Rim on April 18, 2023, 12:03:46 PM
I have issues with my HDMI output.

Hi,

AFAIK, Danne build isn't well supported regarding HDMI currently.
You may want to use crop mood build, HDMI is supported there in all presets in all outputs.

Danne

Yes, recommend bilals build in general and port goodies from this build.

Skinny

Quote from: iaburn on April 04, 2023, 04:54:00 PM
Edit: I've been reading some old posts and it seems like ML is using the reading modes already available in Canon's firmware. So unless there is some magic register combination to enable this, I guess this is sadly not possible u_u

I think it can be done but there will be some limitations.. For example it is possible to read the whole sensor at low FPS, so it should be possible to do it but just skip (not record) every two pixels, saving half of the card space.. So if 10/12 bit recording is possible, it is just lsb truncated? Maybe not recording some pixels also could be done.. But the fps will be limited to the sensor speed probably. Which still can be a good option for recording things such as calm scenes, timelapses and so on. Because full 5k recording takes A LOT of card space..

p.s. I don't really know how it all works so maybe I am saying some total nonsense  :D