Author Topic: Danne's crop_rec_4k experiments for EOS M  (Read 372247 times)

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7702
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1900 on: November 05, 2019, 09:37:02 PM »
speaking crop I used my sigma 10-20 f3.5 and 5dIII in 3k mode and got a really nice wideangle.

DeafEyeJedi

  • Hero Member
  • *****
  • Posts: 3413
  • 5D3 | M1 | 7D | 70D | SL1 | M2 | 50D
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1901 on: November 05, 2019, 10:16:09 PM »
...But 18 FPS is good enough for me.

Isn't this why we all enjoy trial and error, right? Glad you're using the full sensor read out @ 18p which is incredible for static shots!  :D

speaking crop I used my sigma 10-20 f3.5 and 5dIII in 3k mode and got a really nice wideangle.

Indeed, that is super nice. Glad I held on to the Tokina 11-20 f2.8 only because of your never-ending revolution breakthroughs!  :-*
5D3.113 | 5D3.123 | EOSM.203 | 7D.203 | 70D.112 | 100D.101 | EOSM2.* | 50D.109

hokinari

  • New to the forum
  • *
  • Posts: 4
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1902 on: November 05, 2019, 11:46:31 PM »
Hi everyone! Sorry for maybe not correct message for this topic but, is there any chance to see preview of mlv files on windows

DeafEyeJedi

  • Hero Member
  • *****
  • Posts: 3413
  • 5D3 | M1 | 7D | 70D | SL1 | M2 | 50D
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1903 on: November 05, 2019, 11:57:35 PM »
...is there any chance to see preview of mlv files on windows

Yes. Please use MLV_App.
5D3.113 | 5D3.123 | EOSM.203 | 7D.203 | 70D.112 | 100D.101 | EOSM2.* | 50D.109

hokinari

  • New to the forum
  • *
  • Posts: 4
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1904 on: November 06, 2019, 06:15:39 PM »
No, I mean see thumbnails, not just white icons

Walter Schulz

  • Contributor
  • Hero Member
  • *****
  • Posts: 8667
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1905 on: November 06, 2019, 06:26:12 PM »
Requires an MLV codec written for Windows.
Same problem with CR3 in Windows. Canon doesn't provide any either.


Luther

  • Senior
  • ****
  • Posts: 323
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1906 on: November 06, 2019, 06:48:15 PM »
Requires an MLV codec written for Windows.
Same problem with CR3 in Windows. Canon doesn't provide any either.

SageThumbs supports many formats and is open source, although it uses GFL lib, which is proprietary (they could have just used imagemagick or something, don't know why they used GFL). Anyway, Ilia's library could potentially be integrated in one of those thumb viewers for windows...

Walter Schulz

  • Contributor
  • Hero Member
  • *****
  • Posts: 8667
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1907 on: November 07, 2019, 02:21:37 PM »
Thanks for another useful link, Luther!
I leave it to Ilia to decide if it is worth the effort and/or the direction he wants to take. (IANAP)

marodru

  • New to the forum
  • *
  • Posts: 2
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1908 on: November 07, 2019, 04:45:51 PM »
Hello

I have some issues with "2.5k mode". In general it works ok but
 - I have a black bar at the right side of the image so it's like if it is outside of sensor
 - It totally messes HDMI output

My settings are:
 - ML version from Oct 22
 - Preset - 2,5K 2520x1418
 - RAW on - 2420x1054, Aspect ratio in RAW menu - 2.39:1, Preview - Framing (locked), Crop rec preview - Auto. Kill global draw - OFF
 - Aspect ratio in Movie menu - 2.39:1
 - 12 bit depth or 10 bit depth (does not matter)
 - shutter lock at 1/48 with fine tuning (-0.29)
 - FPS override ON at 24fps, Exact FPS priority
 - Sound off
 - Clear overlays - Recording (does not work as its global draw)
 - Advanced screen layout - 16:9 HDMI t/b
 
And sometimes I start having vertical lines over right side of the image, and they GO AWAY if I reboot the camera.

Is that stuff known? What debug data/images/files should I provide?

I have two EOS M cameras now and I'm technical (working in IT for like 30 years) so can participate in testing and data collection pretty deeply.


lightspeed

  • Freshman
  • **
  • Posts: 64
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1909 on: November 08, 2019, 11:15:02 AM »
Would having a black and white only allow greater resolution?

masc

  • Contributor
  • Hero Member
  • *****
  • Posts: 2104
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1910 on: November 08, 2019, 04:30:20 PM »
Would having a black and white only allow greater resolution?
No, it is a bayer sensor, so it is just one channel per pixel.
5D3.113 | EOSM.202

ZEEK

  • Senior
  • ****
  • Posts: 329
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1911 on: November 09, 2019, 12:19:46 AM »
I have some issues with "2.5k mode".
My settings are:
 - shutter lock at 1/48 with fine tuning (-0.29)
 - FPS override ON at 24fps, Exact FPS priority
 - Sound off

*Get rid of these settings for 2.5K Mode (Especially FPS Override). I've been using this 2.5K Mode recently and haven't had these mentioned issues with the EOS M. The monitor preview is perfect as seen on the EOS M Liveview, but "too many" dropped frames to the point it's not useable. With MCM Rewire 1080 these are fine..


EOS M

ilia3101

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1002
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1912 on: November 09, 2019, 12:42:22 PM »
SageThumbs supports many formats and is open source, although it uses GFL lib, which is proprietary (they could have just used imagemagick or something, don't know why they used GFL). Anyway, Ilia's library could potentially be integrated in one of those thumb viewers for windows...

Is that SageThumbs thing easy to write plugins for? I can do it quickly if it's a really simple C api, otherwise I don't want to. I'll just rip out MLV App code and put it in, don't have to wait for LibMLV to be ready. (Done that many times before)

Tried to create a Mac finder thumbnail plugin before and failed, that did traumatise me.

2blackbar

  • Hero Member
  • *****
  • Posts: 514
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1913 on: November 12, 2019, 10:11:22 PM »
Is it possible to center crop modes on my own by editing some reg values ? I need it for 16mm film camera lenses( they should fit in super16mm rectangle, without
 centering they dont) so vignette is not visible in one corner.
Originally crop modes are shifted like this:

And i would like them to be centered like this :

Is that possible ?

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7702
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1914 on: November 12, 2019, 10:20:55 PM »
You can alter width and height to some extent. Also left right. This applies not to all modes. Need more information.
There are also cmos 5 and cmos 1 or 2 I think. All above are included in the crop mode sub menu.

2blackbar

  • Hero Member
  • *****
  • Posts: 514
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1915 on: November 12, 2019, 10:59:31 PM »
Did You found some values that shift image in crop mode up/down or left right ? What regs i need to change and doest it update preview or i need to record it to see it ?
Im in 3x3 mode with set button, changing CMOs 1 and 2 , but without luck,  reg 6800 height and width adjustments also bring no changes to framing preview.

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7702
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1916 on: November 12, 2019, 11:11:07 PM »
Two of the cmos regs moves image to some extent. Try and narrow down which ones. There's info on different places in forum posts. Can't help further atm but if you spend a week or so testing you should be getting what you want.

2blackbar

  • Hero Member
  • *****
  • Posts: 514
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1917 on: November 12, 2019, 11:57:30 PM »
Thanks, i read ML croprec source  for M and found your values, then adjusted them, so for centered 3x3 image i used :
CMOS 5  - 0x380
CMOS 7  - 0xA49
This centers the image as best as possible, CMOS 7 is adjusting up/down so value 48 is too much offcenter and also 50 is too much, not pixel perfect but its best i could do.
Also while experimenting i discovered that it is actually possible to record 10x zoom mode( but i cant reproduce it again) and move around crop window quite freely with ease if you know the values, mostly numeric values work, letters of alphabet wont work but thats not any rule , there are exceptions , but it checks out more often than not.

Is it possible to use these cmos settings for official releases so i dont have to dial them in ?

Ill try to work with 2.5K mode now and center it...
---
Got left/right centered in 2.5k mode :
CMOS(5) 0x2DF
Ill work on up.down
-ok up/down is basically what you already have set in the sourcecode:
CMOS 7 - 0xaa9
So i think thats pretty much it.
I didnt tried if anamorphic is offset yet

Its quite fun process , and not that hard if you know what values you have to adjust slightly
Compared old framing vs new :


Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7702
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1918 on: November 13, 2019, 07:38:01 AM »
Great. I have been tweaking those numbers a lot bit refinements are welcome.
To exclude implementation of wrong presets I need exact name of preset and if using x3zoom mode or not.

2blackbar

  • Hero Member
  • *****
  • Posts: 514
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1919 on: November 13, 2019, 08:25:45 AM »
First one cmos5 - 380 and cmos7 - a49 is for 3x3 crop mode, entered using set button in mcm rewire and started adjustments to cmos 5 and 7.
Second values cmos5 - 2df and cmos7 - aa9 are for 2.5k mode 10 bit ( i switch to 12 bit cause impact is minimal and with 10bit you get banding in darks like we used to get in old 8 bit raw mode)from selector menu and it works in 16:9 and cinemascope 2.39:1 aspect ratios
I tried to refine them more but image gets squeezed with some values and when i go for higher values to desqueeze then framing moves too much to opposite side already.
All done in 23.976 fps.recorded and tested.

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7702
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1920 on: November 13, 2019, 09:29:21 AM »
First one cmos5 - 380 and cmos7 - a49 is for 3x3 crop mode, entered using set button in mcm rewire and started adjustments to cmos 5 and 7.
Second values cmos5 - 2df and cmos7 - aa9 are for 2.5k mode 10 bit ( i switch to 12 bit cause impact is minimal and with 10bit you get banding in darks like we used to get in old 8 bit raw mode)from selector menu and it works in 16:9 and cinemascope 2.39:1 aspect ratios
I tried to refine them more but image gets squeezed with some values and when i go for higher values to desqueeze then framing moves too much to opposite side already.
All done in 23.976 fps.recorded and tested.

Still very cluttery. Do you mean:
mv1080p MCM rewire in x3 zoom mode cmos5=380, cmos7=a49
2.5K 1:1 centered cmos5=2df, cmos7=aa9

2blackbar

  • Hero Member
  • *****
  • Posts: 514
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1921 on: November 13, 2019, 09:46:12 AM »
Yes i suspect this
Code: [Select]
             
           if (x3crop == 0x1 || crop_patch)
                {
                    cmos_new[5] = 0x380;
                    cmos_new[7] = 0xa6a;
                }
Could be made into this :
Code: [Select]
             
           if (x3crop == 0x1 || crop_patch)
                {
                    cmos_new[5] = 0x380;
                    cmos_new[7] = 0xa49;
                }

and this :
Code: [Select]
  case CROP_PRESET_3K_EOSM:
                if (get_halfshutter_pressed() && gain_buttons && !RECORDING && is_movie_mode())
                {
                    return;
                }
                cmos_new[5] = 0x280;             /* vertical (first|last) */
                cmos_new[7] = 0xaa9;            /* horizontal offset (mask 0xFF0) */

to this :
Code: [Select]
  case CROP_PRESET_3K_EOSM:
                if (get_halfshutter_pressed() && gain_buttons && !RECORDING && is_movie_mode())
                {
                    return;
                }
                cmos_new[5] = 0x2df;             /* vertical (first|last) */
                cmos_new[7] = 0xaa9;            /* horizontal offset (mask 0xFF0) */

up/down horizontal framing was already fine, left/right vertical was a bit off.


Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7702
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1922 on: November 13, 2019, 09:51:04 AM »
I'll look into it. Thanks.

2blackbar

  • Hero Member
  • *****
  • Posts: 514
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1923 on: November 13, 2019, 10:22:49 AM »
I found centered anamorphic 5k ( anamorphic 5k set from jhop selector) cinemascope 2.39:1 ( when switching to 16:9 it works too)
CMOS7 - 0xf27
I also made this reference image for different sensors , if anyone would like to tinker with CMOS 5 and 7, its easy just display this image on monitor and zoomin to apsc or fullframe depends on your camera, then enter crop modes and adjust, you can already find some CMOS values in sourcecode .
http://ipic.su/img/img7/fs/sensors.1573634342.png

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7702
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1924 on: November 13, 2019, 12:13:02 PM »
What's jhop selector?
Not working with anamorphic rewired mode?