Messages

If it is real: how much do we know about the usb interface chip?
It would be great to get something like adb working to have an interactive shell.

1. April? :/

@Renovatio: I forgot about the max. crop buffer size, which is 2000x1080 for the 50D - so yes if you set 2048 it will fall back to the next multiple of 32 below 2000, which is 1984. So 1984 is indeed the achievable maximum.

Well the issue is that the next higher value you can set in mlv_rec after 1920 is 2048 - which is too much for the card interface.

Here the max. achievable frame sizes in crop mode - if the card plays along:
2.39:1    1920 x  804  24p     61.8 MB/s     4.01x eff. crop
2.35:1    1920 x  818  24p     62.9 MB/s     4.01x eff. crop
1.85:1    1792 x  968  24p     69.5 MB/s     4.30x eff. crop
16:9       1728 x  972  24p     67.3 MB/s     4.46x eff. crop

2000x852 is 2.35 not 2.39.

Also the x resolution must be a multiple of 32.

Depending on the framerate one should be able to achieve the following in crop mode (assuming max 70MB/s and using current mlv_rec):

• 2.39:1    1920 x 804  24p    61.8 MB/s    4.01x  eff. crop
• 2.39:1    1792 x 750  30p    67.3 MB/s    4.30x  eff. crop

     

DaVinci Resolve needs a screen resolution of at least 1920x1080, it sounds like your monitor doesn't meet that requirement.

Two things:
1) when you shoot mlv/raw your flat profile doesn't have any effect at all.
2) You can have a completely free workflow by using Davinci Resolve Lite, it can read in DNG sequences, no need to convert to TIFF files, you can color correct etc and you can export to h264.

Actually with the EOS-M in raw non-crop mode it only records a vertically squeezed image due to 3x5 line skipping e.g. it records every 3rd column but only every 5th row and such the image has to be unsqueezed by 5/3=1.666 to get the correct ratio again.

On EOS-M only the magic zoom/crop mode gives raw that doesn't need to be streched - even the 1080p modes. Also recording 60/50fps raw doesn't work.

Well, dual iso managed to save 'DUAL2086.CR2' successfully... so that worked.

EOS-M:

I compiled the autoexec.bin myself from the recovery branch using the portable platform.

Interestingly when one compares the output for my EOS-M with the one from jpaana - the second IMG-naming property looks correctly resolved on mine but not on his? I have the feeling something with the property parsing is not going correctly.

Mail with a short sample is on the way.

I tried converting a mlv from a 500D - looking at your code I would assume a colormatrix for the 500D is missing.

Excerpts from the log:

Code Select Expand


------------- 1.7.0 started at 15.03.09 08:18:57 --
[init][settings] -- ffmpeg Exists: false

[createMLVBlocklist] indexing Blocks in C:\Users\Peter Banane\Desktop\video\janarashtest\M09-0943.MLV
[getMLVAttrib] started
[getMLVAttrib] reading RAWI-Block
[getMLVAttrib] reading MLVI-Block
[getMLVAttrib] reading IDNT-Block
[getMLVAttrib] [!] modelName unknown - estimating from colormtrix
[getMLVAttrib] reading WBAL-Block
[getMLVAttrib] reading EXPO-Block
[getMLVAttrib] reading LENS-Block
[getMLVAttrib] reading WAVI-Block
[getMLVAttrib] [!] NO WAVI Block
***** [getMLVAttributes] metaData Object:
xResolution=1536
yResolution=864
frames=65
bitsperSample=14

colorMatrixA=System.Byte[]
colorMatrixB=
forwardMatrixA=
forwardMatrixB=

modell=Canon EOS 500D
camId=MAGIC29cc4ec
apiVersion=0