Author Topic: Danne's crop_rec_4k experiments for EOS M  (Read 240265 times)

Larouso

  • New to the forum
  • *
  • Posts: 14
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #875 on: April 25, 2019, 08:57:28 PM »
Hi all, hope I did not miss it... I skipped some builds... is it on purpose that SDoverclock.lua script is not included in the latest build?


Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7393
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #876 on: April 25, 2019, 09:05:06 PM »
Yes. Sd overclock patch is applied on start up once sd_uhs module is enabled. No need for the script solution.

Larouso

  • New to the forum
  • *
  • Posts: 14
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #877 on: April 25, 2019, 09:10:15 PM »
Thanks !

ErickTim

  • New to the forum
  • *
  • Posts: 2
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #878 on: April 29, 2019, 12:20:42 PM »
Guys... Danne is working really hard to make us awesome new builds, I don't think he has the time to answers questions that had been answered many times before. Someone made this Facebook page where many people will be very happy to answer your questions. This forum is here to help ML team make new builds and continue improving the builds. If you're new to ML I suggest you post your questions in this Facebook page. I myself will answer your questions.

https://www.facebook.com/groups/375018026655820/

Walter Schulz

  • Contributor
  • Hero Member
  • *****
  • Posts: 8488
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #879 on: April 29, 2019, 12:33:45 PM »
Sorry? Where is the benefit in using a second platform (media splitt) hosted by a company with dubious privacy practices which will require another account?
Unasked service, clickbait or what? Not going there and asking forum members to avoid it!

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7393
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #880 on: April 29, 2019, 12:50:38 PM »
I would agree with Walter here. Let´s keep questions and feedback all in one place and one thread. For this very reason haven´t split my branch and build from this eosm thread. Feel free to answer questions in here @ErickTim, and by all means, keep the facebook group but for other reasons than posting about bugs and such.

Levas

  • Contributor
  • Hero Member
  • *****
  • Posts: 1741
  • 6d - Nightly build user
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #881 on: April 29, 2019, 01:20:01 PM »
Also here is interesting experiment using b-spline resize on original dng to 4464:

Looks like using b-spline gets rid of more staircases but perceived sharpness isnt as good.It is interesting option to have IMO.You could then sharpen in post with less aliasing.

@2blackbar: Interesting, the example looks much better with b-spline, how exactly did you do b-spline on the DNG, what software what picture formats etc?

@Danne: Looking at the examples, are you sure horizontal pixelbinning is done in this 1x3 mode ?
On the 6d I use this part for 1x3 mode for pixelbinning:
Code: [Select]
            case CROP_PRESET_1x3:
            case CROP_PRESET_1x3_24FPS:
            adtg_new[11] = (struct adtg_new) {6, 0x8000, 6};    // 6=pixelbinning (8183&8184 needed) 5 is not pixelbinning.
            adtg_new[12] = (struct adtg_new) {6, 0x800C, 0};    // lineskipping
            adtg_new[13] = (struct adtg_new) {6, 0x8183, 0x35}; //sets horizontal pixelbinning, needed if pixelbinning is wanted in 5x zoom mode
            adtg_new[14] = (struct adtg_new) {6, 0x8184, 0x3b};

Because I use 1x3 mode in 5x zoom mode, I need to set 8000 register to value 6 and I have to set values for 8183 and 8184 registers (default values Canon uses in normal non zoom mode)
The difference between horizontal binning or not is very big, both pictures are from 1x3 mode on 6d, the left does horizontal pixelbinning, the right one not.


Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7393
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #882 on: April 29, 2019, 01:25:15 PM »
@Levas
Yes, noticed the jagged edges posted so I double checked my presets and it was fine in my anamorphic modes. No idea why and how he got those jagged lines. Maybe from some early build.

baladev

  • Freshman
  • **
  • Posts: 81
  • 5Dm3 - 6D - 7D - EOS M
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #883 on: April 29, 2019, 02:06:26 PM »
I second that. I've never seen jagged lines like this produced by anamorphic  presets in Danne's builds.. but then again I've never pushed them above and beyond like some people here, who try to squeeze every last pixel and frame out of this camera, sometimes getting weird results like this..

masc

  • Contributor
  • Hero Member
  • *****
  • Posts: 2011
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #884 on: April 29, 2019, 03:19:10 PM »
I think the pictures of 2blackbar are zoomed to 200% or something. With the EOS M I get something like the left picture from Levas: if you look on some ~45° edges, you see the stairs. If you sharpen before resizing you'll get "strong(er) stairs". I also tried with b-spline, lanzcos and bilinear rescaling: bilinear brings the strongest stairs, lanzcos and b-spline looks very close - lanzcos maybe a little sharper, but both smooth the stairs very well.
5D3.113 | EOSM.202

Levas

  • Contributor
  • Hero Member
  • *****
  • Posts: 1741
  • 6d - Nightly build user
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #885 on: April 29, 2019, 04:30:09 PM »
Maybe zoomed to 200% or more indeed ?

The aliasing 'stairs' at 45 degree angles are the only thing spoiling high resolution 1x3 stretch for me.
Did some testing, but couldn't get any better than what I got in the left picture.

arthur2064

  • Just arrived
  • *
  • Posts: 1
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #886 on: April 30, 2019, 11:07:13 AM »
Good morning, for some time I wanted record videos and I opted for the camera eos m, I want to install magic lantern but I am confused about what version to use?. I read that dannephoto versions are highly recommended but I do not know what would be the best for me, I just want to record at 2.5k, Anamorphic Module, 14bit lossles and sd hack module , please guide me.

Also i could not find information for field monitor compatible with ml, does anyone have any idea which ones are compatible? Thank you very much.
PS: sorry for my bad english.

masc

  • Contributor
  • Hero Member
  • *****
  • Posts: 2011
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #887 on: April 30, 2019, 02:46:32 PM »
... I just want to record at 2.5k, Anamorphic Module, 14bit lossles and sd hack module...
There is exactly one branch: Danne's experiment branch. https://www.magiclantern.fm/forum/index.php?topic=9741.msg208959#msg208959
5D3.113 | EOSM.202

Luther

  • Senior
  • ****
  • Posts: 323
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #888 on: April 30, 2019, 03:20:10 PM »
There is exactly one branch: Danne's experiment branch. https://www.magiclantern.fm/forum/index.php?topic=9741.msg208959#msg208959

Wouldn't it be better to push at least some of those changes into main tree? Lot's of questions in this thread this month about where to find the right package for EOS M. Is it too unstable yet?

baladev

  • Freshman
  • **
  • Posts: 81
  • 5Dm3 - 6D - 7D - EOS M
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #889 on: April 30, 2019, 04:23:58 PM »
I agree, it would have been much simpler for everyone if the first post in a thread dedicated to a particular camera build was where the description of the build with the link was. It is standard practice on most forums dedicated to releasing mods and such. It would reduce number of questions like above.

In terms of stability, Danne's build is rock solid and has all the features available for this camera. The only problem pretty much is a phenomenon of pink/corrupted frames that occur in footage produced in crop or anamorphic modes. I'm planning to have a go at it when I finish my current dealings with the 6d build.

masc

  • Contributor
  • Hero Member
  • *****
  • Posts: 2011
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #890 on: April 30, 2019, 05:05:53 PM »
... on the other side: this link can be found on nearly every page in this thread (>= page 71). So reading and finding needs not longer than 3 minutes! I am sorry to not be able to understand the confusion: the link is always the same.
5D3.113 | EOSM.202

2blackbar

  • Hero Member
  • *****
  • Posts: 507
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #891 on: April 30, 2019, 06:09:58 PM »
I used this and compared results
https://www.imageenlarger.com/
These are 100% crop images

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7393
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #892 on: April 30, 2019, 08:03:09 PM »
Hey guys. I wish I had more time. I have a hard time estimating hours put into testing and hacking/understanding parts of ml code so I could make good use of my ideas but atm energy is rather low. All my dedication went into testing out and building a stable version so now I rely on users to point to link/posts giving helping hands when I can't.

Even if working rather stable I see most of my solutions rather primitive and probably could need a real workthrough. Yet all code is there, I even provide compiled builds, so not a big problem finding it. What seems harder is getting newcomers to be more patient and putting in the effort of reading through forum posts. Hey, what can I say. I am rather lazy about that too sometimes.

Still occasional issues with corrupted frames. If we could create areas to dig into regarding those issues maybe we could eliminate these issues. Personally I hit a road block on the subject.

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7393
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #893 on: May 01, 2019, 02:46:10 PM »
New version:
https://www.magiclantern.fm/forum/index.php?topic=9741.msg208959#msg208959

- if x3 zoom is active with mv1080p MCM rewire mode it now gives 1736x1158 for x3 zoom too

Commit:
https://bitbucket.org/Dannephoto/magic-lantern/commits/a89e83d20b77c5d95a1d3fc0a6201bbfdd053343?at=crop_rec_4k_mlv_snd_isogain_1x3_presets


EDIT: uploaded new version just now. Centred the x3 zoom with cmos 5

Larouso

  • New to the forum
  • *
  • Posts: 14
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #894 on: May 01, 2019, 09:24:53 PM »
I had some time today to test the build from 24th April. Always started with cinema script. Worked all good, the only thing that I encountered on regular basis was a completely white screen when start recording (in various crop modes). Video itsself was recorded ok. To me it strikes that it only happens when I set ISO to auto... don`t kow if this already a known thing.

BTW: Is there any way to you use zooming 10x on half shutter release (or any other button) also in video mode like in photo mode (to align fous)?
Have a nice evening all.


Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7393
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #895 on: May 01, 2019, 10:15:34 PM »
As long as footage is ok a white flash in beginning is ok. If not using auto iso, works differently?
What happens when you zoom into x10? I tend to use magic zoom for closer previewing capability.

Larouso

  • New to the forum
  • *
  • Posts: 14
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #896 on: May 01, 2019, 11:37:15 PM »
As long as footage is ok a white flash in beginning is ok. If not using auto iso, works differently?
It`s not a flash - it stays completely white, so you have no preview of what you are capturing.  My assumption was, it is kind of a way overexposed preview / live view.
If auto iso is not enabled I never had the issue.
-> just checked again in a dark environment - it is for sure an overexpose thing - in dark environment you can see slightly the real preview like under a transparent white layer.

What happens when you zoom into x10? I tend to use magic zoom for closer previewing capability.
Yes right... In photo-mode I can use 10x zoom to really nail focus on the subject. In video mode magic zoom is available, true, but with its 3x I found it is not that precise. If you film 1080 and you subject is farer away it`s already pretty hard to determine if you  have the correct focus point. Just was wondering if there is a way to enable that also in video.

dfort

  • Guest
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #897 on: May 02, 2019, 04:51:58 AM »
If you are using Auto Movie Mode with raw video you should see this warning flash on the screen:

Quote
!!! Auto exposure !!!
Set 'Movie Exposure -> Manual' from Canon menu

Heed this warning.

2blackbar

  • Hero Member
  • *****
  • Posts: 507
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #898 on: May 02, 2019, 10:38:16 AM »
Great work Danne ! I think most of us understand why you would like people to read more, You put a lot of effort into Your work and they should put some too, its also great way for them to learn more about ML and why some functions behave certain way.

marcelolippi

  • New to the forum
  • *
  • Posts: 42
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #899 on: May 03, 2019, 10:41:20 PM »
today I started a documentary filmed with the eos m, 150gb, 1480x1900, 23.97, 4.5k, sandisk 128gb extreme pro 170MBs, focus peaking and magic zoom on. Only 2 clips stopped. Maximum duration of one 17 minutes without a pink frame. The final clip fills the card but I still find it legible. Thanks, great work!