Recent Posts

Pages: [1] 2 3 ... 10
1
Camera-specific Development / Re: Canon EOS M
« Last post by loknar on Today at 09:03:57 AM »
It took me while to figure out what "the other routine" means  ???
You were right, with SDR50 it works at 240 MHz @ 73MB/s  :D
Thank you.
2
I've updated a branch that applies twice the focus pixel fix for dual ISO, the second one after the dual ISO processing, like a non-dual ISO video, and the results are much better on my tests. More obvious on binning modes due to larger stretched pixels, but less artifacts in all modes.
A comparison, better download and switch from one to another:

No focus pixels fix:


Normal fix, the current one:


Double fix:


nice!
3
Reverse Engineering / Re: LiveView Investigation
« Last post by vastunghia on Today at 07:24:49 AM »
improvement for 5d3 will be a little revolution because 5D3 is far better in terme of quality. It will not only be a cheap camera to achieve raw footages but a real cinematique camera for professionnal work.

+1 and I also think that preview improvements over HDMI, in particular, will be crucial.
4
Ok, my bad. I thought your clip was from 5k anamorphic frtp and ratio set to off.

That particular preset I missed to fine tune.
In general I will focus mainly on the top presets in the tap display list. Bottom ones, not so much, used too little.
You can easily spot if output looks wrong now I guess.
5
"Pay particular attention to M06-1306"
If you change fps to 20 fps, what do you expect to happen? Also height is altered?
My fix was about finetuning fps so your modification breaks the fix. It is very easy to see what happens by viewing the scanlines in dual iso output. In your example the lines are creeping around, if you look at the other clips output are showing calm, static lines.
If you intend to alter fps, at least find sweet spot and achieve static preview on your camera before recording.
In HD1080p dualiso won´t be flicker free as it runs native canon fps. Might fix this but dualiso with 3x3 looks like garbage mostly anyway.

Added 1736x2180 pixel map and uploaded an eosm build with the new resolution.

Danne, maybe you should pay attention to your own stuff! Where did you get all that stuff you are talking about here? You asked me to send a similar clip at 14 bits, that's what I did!
Your preset defaults to 20 fps. Like I said, I used the build stock and only changed the ratio to 2.39:1...

Edit: I just double checked. All the 1736x2180 clips recorded in that batch recorded at the same  20.005 fps rate, and the 2520x1054 clips all recorded at 23.976 fps. So it's not a glitch.
6
@theBilalFakhouri

    Ok I understand. Thank you for the quick reply. I'll have to check out Discord, thanks for the info.

- Paul
7
"Pay particular attention to M06-1306"
If you change fps to 20 fps, what do you expect to happen? Also height is altered?
My fix was about finetuning fps so your modification breaks the fix. It is very easy to see what happens by viewing the scanlines in dual iso output. In your example the lines are creeping around, if you look at the other clips output are showing calm, static lines.
If you intend to alter fps, at least find sweet spot and achieve static preview on your camera before recording.
In HD1080p dualiso won´t be flicker free as it runs native canon fps. Might fix this but dualiso with 3x3 looks like garbage mostly anyway.

Added 1736x2180 pixel map and uploaded an eosm build with the new resolution.

8
Here is another Batch of clips, mostly 14bits like you requested: http://bit.ly/3X7hATq

Clips were all recorded with Histogram showing 0.5.

Pay particular attention to M06-1306. I included a MLVApp receipt which will emphasize the point of interest.

One has to be careful with exposure, as when in dual Iso mode,  False colors and Histogram are not spot on (Those register a lot more headroom than what is actually available). 14bits is more forgiving, but I often clipped the snow outside when monitoring with those. I got good recordings when maximizing at both the base and recovering ISO individually, in single ISO mode and then enabled dualISO.

If you look at the waveform monitor in MlvApp, you can sometimes see that the absolute level is below clipping, but the shape of the drawing (Flat Top) suggests otherwise.


Recorded with 2023Feb06 Build.
9
All of the provided clips looks good when processed. If we get back to this clip:
M04-1236-BldFeb042023am.MLV it will give a subtle flicker. It´s 10bit though. Is it possible to record a similar clip at 14bit? Dual iso is strongest when 14bit. 10bit not so useful imo.

Here is another Batch of clips, mostly 14bits like you requested: http://bit.ly/3X7hATq

Clips were all recorded with Histogram showing 0.5.

Pay particular attention to M06-1306. I included a MLVApp receipt which will emphasize the point of interest.

One has to be careful with exposure, as when in dual Iso mode,  False colors and Histogram are not spot on (Those register a lot more headroom than what is actually available). 14bits is more forgiving, but I often clipped the snow outside when monitoring with those. I got good recordings when maximizing at both the base and recovering ISO individually, in single ISO mode and then enabled dualISO.


Recorded with 2023Feb06 Build.
10
Camera-specific Development / Re: Canon EOS M
« Last post by Danne on Yesterday at 11:21:59 PM »
Yes but could work with 240 if patched with the other routine.
Pages: [1] 2 3 ... 10