Messages

Yikes! But why is your CF whitebalance right and the raw2dng opened in ACR totally wrong?
Is there some camera wb-data transfered in your process? Maybe I should try set the wrong WB in camera and see if that wrong get's copied to CF.

Yep but you do write something I guess? Why else is the 5D-CF file correct and BMCC totally green?

I just played with my BMC files once more. I have to transfer the RED/GREEN/BLUE values from the whitebalance panel in Cineform Studio to resolve to get a clean BMCC file.

The 5D-CF file loads just fine in Resolve, I have no idea why, David may know. And about the quality of that BMCC-debayer I'll continue in the other forum

Yep, just download it and see for yourself
As long as you export only 1920x1080 and don't need the Resolve-NR it's free.

Too bad the raw control is not implemented with cineform. With DNG you can set exposure/wb/shift, with R3D you can set a lot in Resolve directly
Really hoping to find a way to change wb/curve, I wonder how David's converter set's the WB from the .RAW

The cineform sample has some color artifacts; was it debayered by the cineform software, or with the Resolve algorithm? If by CineForm, which debayering method did you use?

It's the same .mov! Resolve normally totally ignores the Cineform Studio processing and reads pure raw data. With that you get this flat but high quality image. But there is no way to change output curve or white balance.
The 3rd picture is with David's Resolve patch. What that patch does is force Resolve to act like any other windows application meaning get the debayered processed image from CineformStudio. All metadata settings in Cineform are shown this way and I used advanced detail 2 as debayer algorithm.

So the best quality would be to go via Resolve unfortunately I don't know how to change whitebalance/curve with that. That's my question to David, is there a way to edit the .mov because there IS some wb-information stored in the file why else would it be fairly accurate and not totally of

Ok, did some more digging with the patched and unpatched version for Resolve.
Remember the green cast on the unpatched version David? It's apparently a BMCC only problem. The 5D-CF-RAW files open just fine in Resolve in the unpatched version. They open with very little contrast I expect this to be the decoding curve I set when converting. Unfortunately there is no way to set the whitebalance as Studio changes are completely ignored BUT the whitebalance seems right, are you reading it from the .RAW file? It looks ok whereas the DNG in ACR always have that magenta shift?

Ok now for the quality comparsion. Resolve works lovely
As it opened very flat I added some contrast, thats all I did in Resolve, if you compare it to the ACR I opened with sharpening 0 and noisereduction 0 they almost look absolutely identical! I then installed the patch, chose ADV Detail 2 and reexported (with no contrast adjustements) from Resolve. Too bad there is no way to set the whitebalance/decoding curve for Resolve. Is there maybe a way to modify the .mov file instead of doing just metadata via Studio?





And well the BMCC images are completely of, but wrong topic for this forum

Yes that pink highlight restauration is ACR magic.

@Samuel: I was surprised by the difference too, the BMC is definately sharper in that test. But keep in mind that those are very very fine lines and small fonts so in real world the .RAW is just fine

Ahh I forgot. Ok let's assume I run the version with no patch, how can I set whitebalance and basis settings in Resolve to get rid of that green cast?

ehm Samuel wait, I did not run cineform on that test. I did CR2 (still image), RAW (via ACR) and H264. To be honest I don't remember the sharpening settings but I think it was just the default 25 with radius 1.0 - For the CR2 I used resize sharper in photoshop to resize to HD resolution.

Isn't CF advance detail 2 the best option for good sharpnes with no excessive artifacts David?

Is Resolve doing the demosaic not cineform? So no matter what I select in GoPro Studio Resolve handels it? Or does this behaviour change since I use your patched dll?

Take a look at my comparsion pics Samuel. It may not be as sharp as ACR but no way it's as mushy as the H264

https://plus.google.com/u/0/b/111134722005606605004/111134722005606605004/posts/TE4uJBWZU5m

BTW I just opened the image for comparsion in Photoshop. As nice as Cineform is, the Adobe Denoiser does a great job on the bookshelf when lifting the shadows ...

  So this should work -(8,0,1928,960) -- except I introduced a bug it doesn't.  Fix coming.

Works perfectly, thanks alot! Imho this is the way to film with the 5D3 (selecting 2048 it jumps to 1928) because otherwise you'll always have the 2 black rows on the left.

I did have a look at the BMDFilm output curve, and it is a little odd.  It is likely has BMD sensor calibration stuff in it, the curve we use for Protune and others, mathematically models the mapping of linear to log and back again, the BMD curve don't fit our models.

Oh ok didn't know that. I found a thread on bmcuser where somebody found a simulation using the Cineon 95 curve. I did some further testing and pushing with some frames of mine and found a pretty matching preset. BMDfilm compresses/lifts the shadows a little bit more but otherwise it fit's good imho. (Here's the link with the BMCC comparsion: http://www.bmcuser.com/showthread.php?2306-Cineform-Raw-BMD-Film-Curve&p=65225&viewfull=1#post65225)

Attached a screenshot of GoProCineformStudio with Protune curve, one with my simulated bmdfilm and one with the bmd to rec.709 lut applied. Yes I'm digging more into the shadows on the bmdfilm like grade and yes whitebalance probably has to get adjusted it was just a quick apply with no correction and my settings were originally for the bmcc.

Already found a bug..,kinda

I always record in 1928 width because we have 2 lines garbage on the left. recording 1928 and cutting 7 pixel on the left and 1 pixel on the right gives exactly the same FOV as a CR2 still. But if I convert this raw with the converter and want to play it back the program crashes. Be it the quicktime player or gopro studio or windows media player when trying an avi.
I tried to crop it via parameter but seem to do it wrong? How are the parameters? is -(7,0,1926,959) correct?

Happy to give it a try David, do you still need some .raw files?

Did you take a look at the BMDfilm output curve? Do you think it would be possible to get that as output so that I can grade in Resolve with a flat image (like the bmdfilm)?

So you can record 1920x1080 with no dropped frames on the Transcend 32?

How fast are your 32 in liveview?

I bought the Transcend based on the chart but that is for the 32gb. Still it says write speed 120mb/s on the box. Dunno if my one is just a bad batch or if the 32 really is that much faster then the 64gb

My one unfortunately is not enough for 1920x1080 @24p - even with the 3Dhack it drops 1 frame every ~2 seconds.
Speed is unchanged with or without global draw. In 1928x960 I can get 25p even with global draw on

Uhm just a stupid question. Why is 60p such a strange crop/stretched? If I choose a low resolution and do 25p it takes a crop in the right aspect ratio, if I select 50p it takes all height? Any reason for that?

Ah found the hack3d.

With that enabled in 1920x1080 it skipps a frame about every 2 seconds

Have you tried turning on Hak3d in the raw menu?

Hak3D?

--> when running benchmark from photomode i get depending on buffer ~100
--> in videomode i get 85-95mb/s depending on buffer, 95 would be enough but I don't know what kind of buffer is used

Unfortunately I can't get 24p working in full 1920x1080 mode on my 1000x Transcend 64gb. I posted some benchmarks in the other thread. If I read the already posted chart correctly the 32gb transcend seems to be faster. In 1920x1080 I have about 1 frame skipped every 1-2 seconds Just a tad too slow

One more time Transcend 64GB 1000x - video mode - this time with HDMI enabled (even less performance)

Code Select Expand

Buffer size experiment
ML v2.3.NEXT.2013May19.5D3113,
Mode: movie 1920x1080 25p, Global Draw: OFF
CF TRANSCEND        TS64GCF1000                             
10312704 692
15885312 689
80896 205
8949760 726
28661760 737
16264192 711
24521728 720
16424960 674
13148160 690
15619072 679
1658880 585
21001216 776
8548352 696
19760128 772
5965824 743
24906752 758
13968384 703
16770048 742
4542464 679
13063168 748
33263616 792
30312448 757
16713728 714
3887104 704
21841920 789
12846080 714
16192512 751
3333120 685
6719488 729
17313792 785
23374848 764
2865152 683
32488448 757
18588672 741
4186112 713
821248 582
3661824 737
18359296 772
19875840 729
22754304 724
13665280 709
24957952 750
28555264 724
33442816 733
10708992 766
20242432 766
8774656 722
11216896 722
23156736 747
12750848 713
12369920 752
2157568 668
15279104 771
29943808 744
30321664 757
19004416 716
268288 389
23778304 789
467968 495
1905664 659
32368640 774
19829760 762
26281984 747
24153088 747
11364352 734
32160768 761
5944320 689
24118272 774
28511232 772
31526912 751
15956992 727
1732608 642
13881344 783
6106112 705

Same card, photo mode:

Code Select Expand

Buffer size experiment
ML v2.3.NEXT.2013May19.5D3113,
Mode: playback, Global Draw: OFF
CF TRANSCEND        TS64GCF1000                             
10312704 899
15885312 930
80896 360
8949760 989
28661760 930
16264192 938
24521728 917
16424960 906
13148160 915
15619072 900
1658880 825
21001216 910
8548352 844
19760128 1013
5965824 1010
24906752 961
13968384 936
16770048 950
4542464 910
13063168 1001
33263616 1052
30312448 1009
16713728 945
3887104 988
21841920 1076
12846080 942
16192512 1023
3333120 963
6719488 993
17313792 1098
23374848 1031
2865152 908
32488448 1050
18588672 1015
4186112 984
821248 918
3661824 1039
18359296 1027
19875840 981
22754304 970
13665280 922
24957952 1009
28555264 955
33442816 960
10708992 970
20242432 987
8774656 954
11216896 918
23156736 961
12750848 927
12369920 996
2157568 988
15279104 1039
29943808 995
30321664 1011
19004416 941
268288 641
23778304 1051
467968 757
1905664 1001
32368640 1052
19829760 993
26281984 1009
24153088 996
11364352 993
32160768 1013
5944320 960
24118272 1030
28511232 1036
31526912 986
15956992 968
1732608 932
13881344 1085
6106112 951
17367040 1051
21693440 1056
30835712 1040
1139712 848
13698048 1092
19388416 967
1532928 940
16333824 1086
15963136 957
23540736 1045
23770112 1007
7317504 970
6019072 1051
13303808 1000
32039936 989
13137920 1004
24311808 1031
25442304 1015
8986624 946
25421824 1045
2720768 908
18526208 977