5D2 RAW video Builds 14-Bit

Started by a.d., May 20, 2013, 05:27:13 PM

Previous topic - Next topic

0 Members and 2 Guests are viewing this topic.

sdesign

uh, yea, less fps= more room in the buffer.

vikado

Quote from: Redrocks on June 27, 2013, 09:07:42 PM
"ps-do we still have to put the camera mode on C1/C2/C3 mode for more buffer?"

Correct me if I'm wrong, but I think the perceived effect of the C modes was down to photo settings and switching them down.
so at this point, C-modes have no bearing on buffer space.
5d2 user

1%

If you kill the timer completely you might get more speed.

C modes should have never had effect on buffer... but you probably had different settings in them which did.

I don't see why 5DII can't do 80MB/s when 50D can.

If you no-op this, timer should be off completely:

ROM:FFAAF418                 BL      str_StartDialogRefreshTimer

Doyle4

I noticed under Small Hacks it says, Disable Auto White balance and exposure, does it mean it is disabling it or we have too manually? i know how to disable Auto White balance but not exposure if someone could let me know please.

Cheers.

UPDATED: It is under Live View/Movie func. / Stills+Movies and then be sure to be set to Movie Display. If
it is set to Exposure Simulation, the camera, after hitting the record button, will go to auto exposure.

reddeercity

 :) Hi All,
Just  tested the New Bulit by  A_D_  today June 27th .(b77a85f7ac68)
Great speed , I didn't think the 70MB/s wall would be push back  :o
I have Never had any  pixel issues  before  like other had, so
i tried his hack, & i  am seeing speeds close to 80MB/s Stable ;D
i tested the Frame sizes that i would use & the ones that are Continuous.

1856x1004 24p 1:85.1  74.7-75.1MB/s Continuous.
1856x1044 24p 16x9    75.6 MB/s         2300 Fame?
1872x1012 24p 1:85.1  75.1 MB/s        3925 Frames
1872x940 24p   2:1.1    73.4 MB/s        Continuous.
1880x1016 24p 1:85.1  74.7MB/s        2526 Frames
1880x940   24p 2:1.1    71.8-72.3 MB/s    Continuous.
Crop:
2048x930 24p   2:20.1  76.3 MB/s         Continuous
2048x1024 24p 2:1.1    78.1 MB/s         890 Frames
2144x896 24p   2:39.1  77.9 MB/s         Continuous
2144x912 24p   2:35.1   78.2MB/s         5167 Frames

Tested on Lexar1000x 32 GB
Frame over ride set to 24p Exact
Small Hack on
Jpeg small, Every thing on default.


;D

qma5D

Complete newb, but I read the whole post and wasn't able to find anyone with an issue like mine.

I've gotten everything to work except the FPS override.  When I toggle the options it first shows 23.976, then switches to 2.502.  I'm not able to get to either 24 or 25. 

Is there another setting that I'm missing to unlock 24 or 25?  Thanks!

I am using 282abfa87fef from 6/25.

reddeercity

Quote from: qma5D on June 28, 2013, 02:08:36 AM
Complete newb, but I read the whole post and wasn't able to find anyone with an issue like mine.

I've gotten everything to work except the FPS override.  When I toggle the options it first shows 23.976, then switches to 2.502.  I'm not able to get to either 24 or 25. 

Is there another setting that I'm missing to unlock 24 or 25?  Thanks!

I am using 282abfa87fef from 6/25.
Make sure in "frame over ride" you go to
"Optimize For" set to "Exact FPS"
then set to (in my case 24)
make sure in the Canon Live View/Movie func. set. menu
you are set to 1920x1080 24.
   ;)

qma5D

I only have the option for 1080/23.98 or 1080/27.97 in the set menu.

I have 1080/23.98 selected. 

I don't have any of those options in the "Frame Over Ride" section.  I can only select "2.502" or "Off"

Thanks for your response!

ted ramasola

Jun 27 build is an awesome leap in performance.
Amazing job ML team!

Below is a comparison between the best performance of jun 26 with "variable buffer"
to Jun 27 with "small hacks".

Figures separated by a / indicate another test record done.

Also included results for "expected" frames, the figure is the last one shown
at the end of recording.
Hopefully this will help A1ex fix the formula to estimate/predict recorded frames.


In the case of 1872 x 1012, for jun 27 build, the figure expected is very high(8663)
compared to actual recorded frames (4686).

More resolutions are now continuous!


Resolutions   a.d. jun 26    a.d. jun 27    expected
1X Mode
1872 x 1054   447/379      1058/937    1127/972
1872 x 1012   769/710      2749/4686   3709/8663
1872 x 936    2764/2011    continuous    continuous
1872 x 850    continuous    continuous    continuous
1856 x 1044   611             2378        2917
1856 x 928    4891           continuous      continuous
1728 x 972     continuous    continuous   continuous
1472 x 1104    continuous    continuous   continuous

Crop Mode 3X

2144 x 1076     219            371          389
2144 x 1072     178            370          382
2144 x 974      326            947        1014
1920 x 1076     347            1326          1548
1856 x 1044     611            continuous     continuous
1728 x 972     contnuous     continuous     continuous

Great job guys!
5DmkII  / 7D
www.ramasolaproductions.com
Texas

noix222

new build 27/06 amazing!@@ super speed i have just one problem!!! i always have the fps override 24 ON but some files seems to come out with 29.97 in 5x and 29.973 in 1x.... at least is what raw2dng or RAWMagic tells me sometimes... NOT ALL FILES COME LIKE THAT its random. Any ideas ?

edit: ok shooting more videos it seems to get stable 24 frames.... pretty strange :D anyway KIIILLLLING IT guysss :D

Yoshiyuki Blade

Do you guys do FPS override 24.000 fps when the LV mode is set to 24 FPS mode or 30 FPS? I found that I can get a shutter angle closer to 180 degrees in 30p mode and I *think* write speeds are very slightly higher (not sure if placebo). Are there any drawbacks with doing 30 FPS mode overridden to 24.000 FPS versus overriding the default 24 FPS mode?

ted ramasola

Quote from: Yoshiyuki Blade on June 28, 2013, 06:49:08 AM
Do you guys do FPS override 24.000 fps when the LV mode is set to 24 FPS mode or 30 FPS? I found that I can get a shutter angle closer to 180 degrees in 30p mode and I *think* write speeds are very slightly higher (not sure if placebo). Are there any drawbacks with doing 30 FPS mode overridden to 24.000 FPS versus overriding the default 24 FPS mode?

My setting is 24p in canon menu, FPS override=23.976.  Definitely more frames than when canon menu set to 30p.
5DmkII  / 7D
www.ramasolaproductions.com
Texas

guentergunter

Quote from: ted ramasola on June 27, 2013, 06:56:10 PM
Using? Like in real project? I would not recommend one at this point since development by the ML team is moving so fast.

For testing, I recommend you follow the current builds and draw your own conclusions based on the CF cards you are using.
Well, now that I've seen the raw, I'll probably never want to use anything else. And I have no fear of using it in real projects by now.
5D2 ML RAW + VAF-5D2b: http://vimeo.com/69350650

a1ex

Can you run some card benchmarks (first or second, whichever you want) in playback mode?

(start the benchmark in photo mode with display off)

I'd like to compare the results with the ones from 50D.

a1ex

Another Canon setting that may make a difference: LCD brightness should be set to Manual.

CPU usage of LightMeasureTask in play mode:
with LCD brightness auto: 19.3%
with LCD brightness manual: 0.1%

andrus

Quote from: a.d. on June 27, 2013, 07:34:57 PM
Hmm.. old versions of CS seem not support opcode
@andrus
Version from 2013/06/08
5D2_WIN_raw2dng.zip
a.d. thank you very much, I can import in AE CS4 with Camera Raw 5.0 with this version, quality is crystal clear!!!
5D MkII | Sigma 50mm f1.4 | Komputerbay 32Gb 1000x | A spare battery :D

Yoshiyuki Blade

@a1ex

Here are the results from a Sandisk Extreme Pro 600x 16GB:


Using a.d.'s build from commit b77a85f7ac68 (6-27-2013).

mgm

June 27 build : Hack On : LCD Manual : Lexar 1000x 32GB

1880x940 @ 73.5MB/s = Continuous
1880x1058 @ 75.7 MB/s = 50 s = 1213 f
1856x1044 @74.3 MB.s = 1:07 s = 1635 f

Incredible speed increase with the hack, - thank you ML team!

bnvm

Here is a benchmark with a Pro Master 1000X 16 GB, June 27th build. small hacks on, LCD set to manual, 23.976 fps.

I am getting about 72.9 MB/s now in crop, 71.2 MB/s in normal. I was maxing out previously at around 63 MB/s - 65 MB/s.

1872 x 1054 = 565 f
1872 x 1012 = 890 f
1872 x 936   = continuous

1856 x 1044 = 789 f
1856 x 1004 = 1505 f
1856 x 928   = continuous

1920 x 1076 = 465 f
1920 x 1038 = 711 f
1920 x 960   = 3461 f
1920 x 818 = continuous

2048 x 930 = 1293 f
2048 x 872 = continuous

What is the best procedure for setting up a benchmark anyway. I just turn on the LV and go but I noticed that sets a mode of Movie... where most test show a mode of photo, this is for the benchmark in Debug BTW. Thanks.

togg


danielcreed

incredible performance leap with the 27th build.

I have a 8Gb Sandisk Extreme at 60mb/s and funny enough i get exactly 60mb/s :)) and sometimes just a bit over.

will test out the resolutions right now.

ted ramasola

Quote from: andrus on June 28, 2013, 02:15:05 PM
a.d. thank you very much, I can import in AE CS4 with Camera Raw 5.0 with this version, quality is crystal clear!!!

andrus,

does that version of raw2dng work with files more than 4gig?
5DmkII  / 7D
www.ramasolaproductions.com
Texas

ted ramasola

Quote from: a.d. on June 27, 2013, 07:34:57 PM
Hmm.. old versions of CS seem not support opcode
@andrus
Version from 2013/06/08
5D2_WIN_raw2dng.zip

@a.d.

does this old version still have issues with the vertical banding noise? I recall the raw2dng was tweaked to fixed the vertical noise issues.
5DmkII  / 7D
www.ramasolaproductions.com
Texas

andrus

Quote from: ted ramasola on June 28, 2013, 08:46:21 PM
andrus,

does that version of raw2dng work with files more than 4gig?

Ted I don't know, I'll have to try and let you know.
5D MkII | Sigma 50mm f1.4 | Komputerbay 32Gb 1000x | A spare battery :D

ted ramasola

Quote from: andrus on June 28, 2013, 09:25:13 PM
Ted I don't know, I'll have to try and let you know.


Andrus,

I just rechecked my files and I do believe it does since the +4gig plus feature was implemented starting with May 21 build for raw2dng.

So jun 8 should be ok, what you should check for instead is vertical banding/noise. I think a1ex and others were tweaking the raw2dng to fix it.

5DmkII  / 7D
www.ramasolaproductions.com
Texas