Author Topic: Danne's crop_rec_4k experiments for EOS M  (Read 241873 times)

DeafEyeJedi

  • Hero Member
  • *****
  • Posts: 3411
  • 5D3 | M1 | 7D | 70D | SL1 | M2 | 50D
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1975 on: November 23, 2019, 02:26:13 AM »
...let ML menu disappear when were changing CMOS values, so we could see how its changing in realtime , without having to exit menu and reenter again to change values.Kinda like how expo menu works , it dosappears when you change iso from ML menu and only iso value stays onscreen.

Great idea, @2blackbar and I'm all for this!

Because im loving how MLVApp outputs image, looks like film straight away , theres something to tonemapped function that made me stop using other software for CR2's when i compared, plain and simple.

I'm on the same boat. Still can't believe I'm actually ditchin' Adobe for photography stuff as well!

That was it !  :D :D :D :D Works, its 4800 at startup( i dont like 5000 or 5200 too warm)

Booyah!!!  :D
5D3.113 | 5D3.123 | EOSM.203 | 7D.203 | 70D.112 | 100D.101 | EOSM2.* | 50D.109

2blackbar

  • Hero Member
  • *****
  • Posts: 507
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1976 on: November 23, 2019, 02:36:43 AM »
Wow DeafEyeJedi man, are those all Your cams? Haha , nice collection.I have only 5D2 and M, but i mostly use M, for FF you need very good lenses without suspicious corners and i dont want to spend a fortune when im not making $ with my cams, thats my rule.
I might be wrong but i think this hides ML menu
Code: [Select]
    if (!menu_active_but_hidden())
    {
       
        int Tv = APEX_TV(lens_info.raw_shutter) * 10/8;
        if (lens_info.raw_shutter) MENU_SET_RINFO(
            "Tv%s%d.%d",
            FMT_FIXEDPOINT1(Tv)
        );
    }
Hidin ML menu during CMOS tinkering is like the first thing i would do ,then exporting addresses to txt file on shutdown

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7395
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1977 on: November 23, 2019, 02:59:12 AM »
Question: why not just use HDRMerge? Don't know if the averaging results are the same, but HDRMerge gives great results already... and not only it does averaging, but also gives you a extreme dynamic range output.
I like the idea of having more open source photo editing tools with new features, and you folks know that me and many others appreciate your work, but sometimes it feels a bit like "reinventing the wheel" or "NIH syndrome".
Big words from someone that doesn't contribute with code, I know. I mean no offense here. I should just shut up.
I tested hdrmerge in a lot of ways. Even built a dedicated compiler for this program. It's not really relevant comparing the averaging workflow outlined so I'd say if you don't see the use simply go on without using it.
Cons with hdrmerge regarding other uses. Output is dng deflate. Tonemapping works on per frame basis(very sensitive frome scene to scene). Hdr movie sequences will not work here for instance. Test go from a dark to bright scene, output will flicker heavily.
Numerous occasions where hdrmerge will output badly merged files. Occasionally big issues in lows(crushing blacks), before highlights were messy too but seems fixed. Sidenote. Tried replacing enfuse merging cr2(well tiff intermediate) files only with a hdrmerge workflow but results were never as reliable as with enfuse. Bulk work not reliable.
Conclusion. For most cases this open source tool works mostly ok, even can give stellar results, but there's more to wish for, format and stability wise.
Feel free to test out hdrmerge, ffmpeg averaging and enfuse in my program Switch. I include a beasty bash script the compares all three of them. Just for fun. Clear winner as always, enfuse. By far.
I think your last words in your post pretty much sums it up pretty perfectly ;).

2blackbar

  • Hero Member
  • *****
  • Posts: 507
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1978 on: November 23, 2019, 03:44:10 AM »
Looks like my Avast antivirus deleted some mingw libraries from cygwin folder... wow...
Have to reinstall again.Dont recommend that piece of shitware , previously i had avira which probably interfered as well.

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7395
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1979 on: November 23, 2019, 03:55:03 AM »
Looks like my Avast antivirus deleted some mingw libraries from cygwin folder... wow...
Have to reinstall again.Dont recommend that piece of shitware , previously i had avira which probably interfered as well.
Too bad :(
By the way. Thought you wanted the actual liveviev output to start out in 4800k :)

2blackbar

  • Hero Member
  • *****
  • Posts: 507
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1980 on: November 23, 2019, 03:59:15 AM »
Yeah i wanted it to be 4800 right away after install and it is now, I just recall what are the changes that i do when i run your builds and i mostly modified all to what i wanted, now i would like to get that ML menu hidden when changing CMOS values so i can see in realtime how image is changing...
--
Ok got it up and running again, wow... but this time im ready.
This line saved me, all docutils
Code: [Select]
"%UserProfile%"\Downloads\setup-x86.exe -s http://cygwin.mirrors.pair.com -q -P mingw64-i686-gcc-core,gcc-core,make,mercurial,python27,python27-docutils,python37,python2-docutils,python3-docutils,python2-docutils,perl,wget,unzip,zip,diffutils

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7395
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1981 on: November 23, 2019, 10:35:41 AM »
New version:
https://www.magiclantern.fm/forum/index.php?topic=9741.msg208959#msg208959

- Added a frame burst mode.
Will stop recording when specified amount of frames have been recorded. Good to have control when averaging a file sequence in post. Should work with all presets.






And of course we all know what a little averaging can do(24 frames recorded at iso 100. Developed in Mlv App:






2blackbar

  • Hero Member
  • *****
  • Posts: 507
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1982 on: November 23, 2019, 01:48:30 PM »
Great work, ill test how much is the difference between 24 and 94
--
Ok i did some tests and my conclusions are :
- 24 and 96 and single iso there is quite big difference, i wouldnt really want to use 24.
- With mcm reqwire resolution mlv files with 200/800/3200 come out nice, you can see dark and bright frames but...
- with fullres anamorphic, the time that isos are changing is too fast, mlv gets recorded but none of the frames are brighter, i think in fullres isos need to be changed slower , or just slow them down in all resolutions.in 24fps mode with fullres i got 17 frames and this one recorded different iso brightness nicely.
I would leave just 96 frames mode by default and maybe slowdown iso change , but i admit 96 frames for fullres is overkill for my laptop, memory bumps up with ffmpeg to almost 3gb.

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7395
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1983 on: November 23, 2019, 02:57:57 PM »
Frame burst is single iso only. Should be.
Iso average  only works with frame burst set to off.
Should work exactly the same for all presets.

2blackbar

  • Hero Member
  • *****
  • Posts: 507
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1984 on: November 23, 2019, 03:19:36 PM »
Ah i see now, its 17 frames in fullres and brightness is correct, so i probably did not switch off frame burst previously.
-----------

Im starting to center 2.5k mode, i know what the issue is with correcting all crop modes and framing being offcenter in 16:9 and cinemascope etc.
Its because when you skip left right up down then youre skipping more from one side, so what i did is change that, now it doesnt skip left 40 and right 0, it skips left 20 and right 20, so when you change aspect ratio, it all stays perfectly centered.
Why ? BEcause then you can use the same CMOS addresses for 2.5k and don thave to change them again if you want centering working in different aspect ratios

So far my values for 2.5k are:
Code: [Select]
  case CROP_PRESET_2K_EOSM:
   cmos_new[5] = 0x2c0;             /* vertical (first|last) */
                cmos_new[7] = 0x6ca;    /* pink highlights without this */
                break;


i basically made skipping to be the same from both sides( while respecting actual aspect ratios like You did) , so there is no issue with centering when changing aspect ratios:
Code: [Select]
        case CROP_PRESET_2K_EOSM:
            /* set ratio preset */
            if (ratios == 0x1)
            {
                skip_left       = 36;
                skip_right      = 36;
                skip_top        = 23;
                skip_bottom     = 23;
            }
            if (ratios == 0x2)
            {
                skip_left       = 36;
                skip_right      = 36;
                skip_top        = 14;
                skip_bottom     = 14;
            }
            if (ratios == 0x3)
            {
                skip_left       = 197;
                skip_right      = 197;
                skip_top        = 106;
                skip_bottom     = 106;
            }
            break;
           

This works, i can see in camera that it helps a lot.
I would of do that for all skipping in croprec by default and in rare cases i would not if something would be off in liveview.
At first i would get skipping from all sides, sometimes i see left skipping or down skipping is ommited which i would not do, i would take half of it and add new line to skip that half from opposite side, and avoid skipping just one side.
This would help getting correct CMOS addresses for different modes because skipping will be uniform from opposites sides ,thre is less variables to take care of.

I suspect You skip sides in 2.5k mode when doing cinemascope so it records mode time ? Isnt it better to leave full 2520 ? So its not overkill for card reader.
CAuse there is 2520x1480 which is already highes 2.5k  preset and its 16:9 but you can also choose 16:9 from other place in menu which crops it to lower 2192x1234
--
Now i see that i i think i know whats up, its stays at 950, so you have to skip sides to get to that 2.39:1
Well... maybe not, its working strange maybe its being changed in other places, also im using aspect ratio calculator and ML gives me different resolutions for 2.39:1 than this calculator
https://www.digitalrebellion.com/webapps/aspectcalc
1920x804   and 2520x1054 is 2.39:1   1920:816 and 2520x1072 is 2.35:1 and i wanted that but i see that i can get 2592 pixels width if i turn off skipping sides on 2.39:1 ratio ( i had 2592x950 maximum recorded wgich is 2.7:1) so i wanted 2592x1084 for 2.39:1 or 2592x1104 for 2.35:1 but i dont know if thats possible.
Ah.. that one aspect ratio in menu (above ratio ) gave me 2592 when i skipped 0 on all sides..
Does this mode wirks ?
Code: [Select]
  case 0xC0F06804: return 0x5b90319 + reg_6804_width + (reg_6804_height << 16); // 3000x1432 24fps x5 Mode;With my own CMOS value for 2.5k that aspect ratio in raw video menu with resolution changing above it, with Your value it crops uniformly like it should, maybe because i added CMOS5 as well not just CMOS 7  it has centering all over the place when i adjust aspect ratio and resolution, where i can look it up so its not so unpredictable and center whatever i can ? I set aspect to 4:3 and resolution much lowe so i should not vignette but i do, because it shifts the frame whenever i adjust aspect and resolution, its not trying to skip from both sides at the same time but from one side more than from other.

It is hard at the moment figuring out CMOS values to center crop modes, when i press menu then everything gets shifted anyway, so there are multiple places where framing gets changed.
--
In Your build ISO climb doesnt work in 2.5k mode, 45/50/48fps mode crop and noncrop, works fine in other modes  .

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7395
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1985 on: November 23, 2019, 10:53:23 PM »
In Your build ISO climb doesnt work in 2.5k mode, 45/50/48fps mode crop and noncrop, works fine in other modes  .
Did you turn off iso average setting? Will set you in iso 100. Iso climb work with all presets over here.
Resolutions have been outlined one by one mainly to deliver continuos recording. However, feel free to refine and customize. Doubt I will have much time to throw into my build any time soon but I keep looking ;).

berechiah

  • Freshman
  • **
  • Posts: 84
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1986 on: November 24, 2019, 03:43:46 AM »
Hey guys any one know how to get rid of 3 or 4 red pixels in my raw video

I have a different eos m
And it has some bad pixels
I did the bad pixel fix in mlvfs
But is there an easier way

DeafEyeJedi

  • Hero Member
  • *****
  • Posts: 3411
  • 5D3 | M1 | 7D | 70D | SL1 | M2 | 50D
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1987 on: November 24, 2019, 05:22:16 AM »
Bad pixel fix in MLV_App does wonders, actually!
5D3.113 | 5D3.123 | EOSM.203 | 7D.203 | 70D.112 | 100D.101 | EOSM2.* | 50D.109

bakersdozen

  • Freshman
  • **
  • Posts: 89
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1988 on: November 24, 2019, 11:09:54 AM »
Bad pixel fix in MLV_App does wonders, actually!
Seconded, almost too easy
EOS M + 5D3

berechiah

  • Freshman
  • **
  • Posts: 84
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1989 on: November 24, 2019, 12:48:56 PM »
Thanks true

Look guys thanks a mill
I sold my eos m about 6 months back
And I had a 70d then a 6d then a 5d 3 and last of all a 5d4

An I am back at the most trusted camera I ever owned
Eosm

I have a 24 to 70 f4 and I am loving it

Thanks again

2blackbar

  • Hero Member
  • *****
  • Posts: 507
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1990 on: November 24, 2019, 10:28:42 PM »
I found this works great for cctv lens and mcm rewire crop without any vignetting from sides
Code: [Select]
                 
                             case CROP_PRESET_mcm_mv1080_EOSM:
                cmos_new[5] = 0x20;
                cmos_new[7] = 0x800;
                if (x3crop == 0x1 || crop_patch)
                {
                    cmos_new[5] = 0x380;
                    cmos_new[7] = 0xaaa;
                }
                if (ratios == 0x0 && (x3crop == 0x1 || crop_patch))
                {
                    cmos_new[5] = 0x380;
                    cmos_new[7] = 0xaaa;
                }
                break;
               

I also have this that works fine for 16:9 1920x1080p resolution(dialed down from 2520x1418) in 2.5k mode 
But... I would like to also center image in 2520x1418 but i only see this for 2.5k mode and cmos settings that i centered while being in dialed down 1920x1080 resolution, its still not centered in 2520x1418:
Code: [Select]
            case CROP_PRESET_2K_EOSM:
      cmos_new[5] = 0x2c0;             /* vertical (first|last) */
              cmos_new[7] = 0x6a9;    /* pink highlights without this */
                break;
               
What should i add if i want to have different centering after i dial down resolution to 16:9 -1920x1080 and different when i use 16:9 2520x1418 ?
I also noticed that after recording mlv it goes back to liveview after slight pause and then framing is wrong , i have to enter playback mode, then press exit in corner, when it comes back to liveview then framing is correct, so after recording there is something thats changing CMOS value again messing up framing.
I also discovered that in 2.5k mode you can actually press small triangle arrows to move your framing around like during zoom, but i tried to get it centered when white zoom box is at the center.

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7395
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1991 on: November 25, 2019, 12:09:29 AM »
Thanks. Hope to get a look at this soon.

Luther

  • Senior
  • ****
  • Posts: 323
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1992 on: November 25, 2019, 09:50:35 PM »
Output is dng deflate.
So what? This is a pro not a "con". Preserving it in raw format gives you much more flexibility.
Quote
Hdr movie sequences will not work here for instance.
Neither ffmpeg tmix.
Quote
Numerous occasions where hdrmerge will output badly merged files. Occasionally big issues in lows(crushing blacks), before highlights were messy too but seems fixed.
I use it for years. Never once this happened.
Quote
Bulk work not reliable.
This point I agree.
Quote
Clear winner as always, enfuse. By far.
I'll test it.
Quote
I think your last words in your post pretty much sums it up pretty perfectly ;).
Maybe. But criticism is necessary sometimes.

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7395
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1993 on: November 26, 2019, 12:45:02 AM »
I prefer coders and deeper know how around modifying/refining actual functions menus etc since my own knowledge is very limited in those areas. Unfortunately they don't grow on trees lately, if ever. Jip-Hop took a short but very useful dive but quickly ran out of time and I have very little time atm myself.
Regarding HDRmerge my points are valid to me and in this case, for the averaging workflow in Mlv app, not at all relevant. If you want to use hdrmerge please do. I won't, for now at least.

Disease

  • New to the forum
  • *
  • Posts: 39
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1994 on: November 26, 2019, 10:46:46 AM »
ML is a bit of fun but it is so unreliable. It constantly deactivated the shutter release so I unable to start recording or taking a photograph without restarting sometimes multiple times.

I have many other questions but I don't know if it's worth noting them all here. I would like to know how to record higher resolutions with some stability, or record more FPS than 30. Having the ability to record 50fps at 1080 would be incredible and seems realistic as people are recording 2.5K at 48fps.

The crop/zoom mode that removes the aliasing and moire always deactivates when I restart the camera and I have to go into the menu and reactivate it. Again makes it kinda useless for anything serious.

One more, can I manually control the audio levels without entering into the menu? maybe set up the dial to control the gain?

Many thanks for any help! :-)

2blackbar

  • Hero Member
  • *****
  • Posts: 507
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1995 on: November 26, 2019, 04:44:15 PM »
Sourcecode is out there, its very possible to fix a lot of stuff , everyone can give it a try and Danne is very supportive.

masc

  • Contributor
  • Hero Member
  • *****
  • Posts: 2011
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1996 on: November 26, 2019, 06:20:20 PM »
ML is a bit of fun but it is so unreliable. It constantly deactivated the shutter release so I unable to start recording or taking a photograph without restarting sometimes multiple times.

I have many other questions but I don't know if it's worth noting them all here. I would like to know how to record higher resolutions with some stability, or record more FPS than 30. Having the ability to record 50fps at 1080 would be incredible and seems realistic as people are recording 2.5K at 48fps.

The crop/zoom mode that removes the aliasing and moire always deactivates when I restart the camera and I have to go into the menu and reactivate it. Again makes it kinda useless for anything serious.

Here it is very reliable. I can record in 5K anamorphic mode all over the day without one single corrupted frame, without one single crash. Same for 45fps at a little less than 1080p. 2.5K at 48fps is not possible with EOS M (and will never be): card interface is way too slow - that is simple math.

What's the problem with shutter release? Absolutely don't know what you mean. I can take pictures at any time in picture mode and can record video at any time in video mode - without changing anything in settings.

For zoom mode in 1080p you have to press exactly one button after start.

Don't write ML is useless and unreliable if you don't know what you do, or if you don't know if you're doing it right.
5D3.113 | EOSM.202

Stousen

  • New to the forum
  • *
  • Posts: 43
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1997 on: November 26, 2019, 06:46:12 PM »
Here it is very reliable. I can record in 5K anamorphic mode all over the day without one single corrupted frame, without one single crash. Same for 45fps at a little less than 1080p. 2.5K at 48fps is not possible with EOS M (and will never be): card interface is way too slow - that is simple math.

What's the problem with shutter release? Absolutely don't know what you mean. I can take pictures at any time in picture mode and can record video at any time in video mode - without changing anything in settings.

For zoom mode in 1080p you have to press exactly one button after start.

Don't write ML is useless and unreliable if you don't know what you do, or if you don't know if you're doing it right.

+1

srsa

  • Developer
  • New to the forum
  • *****
  • Posts: 48
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1998 on: November 26, 2019, 08:25:54 PM »
It constantly deactivated the shutter release so I unable to start recording or taking a photograph without restarting sometimes multiple times.
That's probably the "shutter bug". You can read about it here and also search in the print view of the current thread.

Anaconda_

  • New to the forum
  • *
  • Posts: 21
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1999 on: November 27, 2019, 07:49:40 AM »
I can record in 5K anamorphic mode all over the day without one single corrupted frame, without one single crash.

How? I've been trying for the last few days and often get half a pink frame throughout my recordings. I've tried the preset options and run the script thinking it'll set everything up for me.

Also, do you typically shoot in framing or real-time (or mix between both with the INFO button)?