5D2 RAW video Builds 14-Bit

[reddeercity]

Well i though that's what we are doing, discussing this issue .
And from what I read here the community of 5d2 users  would like to keep
The upper resolutions in the new version of the code.
@Ted no that's not a miss print, that's right 1888.

[guentergunter]

Not only will the 1888 exception make the camera less desirable for recording but if one day the devs decide they will no longer support the 5D2 because of exceptions like this, you will be forced to upgrade. That is what I think the essence of ML is. Upgrading existing gear through firmware updates, unlocking and creating features, so you don't have to buy new gear every year.

Future first!

Future first, but that's also true for image quality!

To be honest, there's something I don't get : what the trouble to just add 1888 (if it's not too much complex to add in code) : All people who do not want to post-prod crop still can use 1856, and others can use max resolution ? (perhaps some notice in the readme file to warn about black border on 5D2, and that's all)

Folks, there's a solution everyone would be happy with:


You will always have to convert your footage (*.raw or *.mlv), since there's absolutely no hope, that these formats will be supported by any major NLE. You have to convert it!

So, what's wrong about the idea to crop 1888 to 1880 while converting to DNG? I guess nothing!


Besides: It's way harder to implement such resolution differences into the camera (due to performance issues, etc.). But in post, where you have to convert anyway it's equal if it takes a hundredth second more for each frame!


So, why not do it this way?

[romainmenke]

Was going to fuel the discussion with some real world tests.
But there is no more 1880 res in the latest nightly.
How far do I have to go back to get that option?

I like the latest nightly by the way.

Preferred settings for tests?

[reddeercity]

Oct 24th Raw a.d. Build has 1880, I still use this built as
It is the fastest built to date for the 5d2.
They never being able to match this build since then it being down hill
For performance for the 5d2. FYI I can record 1880x1058 continuous.
So you better download it before they remove all old builds !

[romainmenke]

Still have that one.
I keep an extensive archive of all software I've ever used, so no worries there.

[Kharak]

Oct 24th Raw a.d. Build has 1880, I still use this built as
It is the fastest built to date for the 5d2.
They never being able to match this build since then it being down hill
For performance for the 5d2. FYI I can record 1880x1058 continuous.
So you better download it before they remove all old builds !

1880x1058 ? so 16:9 ?

Can you post what settings you were using? or pm me

[romainmenke]

blind test

tech specs:
latest nightly vs oct 24
1856 vs 1880

lens: 100mm F2.0 (really sharp if you don't know it)
indoor shots at F8
outdoor at F4 (couldn't reach ideal F8)

Iso 100
1/30s
Tripod mounted
Manual focus

Raw_rec => raw2dng => acr (only white balance) saved as 1080p


we'll name them top to bottom:
a1 / a2 (lots of detail far away)
b1/ b2 (dirty pillow test)
c1 /c2 (dark shiny textures)

please let me know which one is 1880

[romainmenke]

full size images

c1

c2

b1

b2

a1

a2

[reddeercity]

1880x1058 ? so 16:9 ?

Can you post what settings you were using? or pm me

yes, 16x9 . I use Small raw for photo in canon menu
Hacked preview ,with white rectangle frame, small hacks on
No overlays of course when recording raw, on a "freshly" formatted card.
I never change formats without formatting the card, I think this makes it
More stable & faster all on Lexar 1000x 64GB, 32GB .

[Kharak]

please let me know which one is 1880

Please let me know which one is 1880

@Reddeer

You mentioned card warm up off in a previous post as you think Lexar do not benefit from this? Still stands?

[ThibaudS]

Hi everybody !

I need your help because I can not start a RAW pack! I succeeded it a few months ago (I can not find the version).
With 63b2f145cb3b version of February 14, when I run the module, it noted me "We Will load" even when I restart the camera. Whatever the module.

I respect the instructions carefully.
Does anyone have any idea where the problem is? Maybe give me a version where the RAW (mlv or not) module are already running.

Thank you for your answer!
Thibaud

[Walter Schulz]

Delete card's ML directory, copy contents of latest build to card (overwriting AUTOEXEC.BIN) and try again.

[xvince1]

To me :

C1 - 1880 (The BG's screw looks sharper to me)

B1 / B2 - can't figure out

A1 - 1880 (upper windows looks sharper to me)

Yes I agree, it's very difficult to be shure and I may be easily wrong, but I stay my point : what the trouble to keep the max resolution available if there's no coding effort ?

[ted ramasola]

@romainmenke

I know the embedded images you posted are relevant to the current discussion, but due to the forum posting rules of having a max 900pixx900, please post links to the images instead or a 100% crop of the image.

Thanks,

And yeah, there is only a very slight difference between 1880 and 1856 based on those images.

[romainmenke]

couldn't figure out how to embed, so i hosted them on my website and posted a link.
Is that what you mean?

edit: just checked the rules.
sorry your right, will keep that in mind in the future.
Do you want me to change them?

[ted ramasola]

couldn't figure out how to embed, so i hosted them on my website and posted a link.
Is that what you mean?

Yes, but what you did is actually "embed" when you use the "Insert image" function. So either you insert a 900x900 image or just copy paste the url of the image directly from where they are originally uploaded.

Like this:
https://fbcdn-sphotos-b-a.akamaihd.net/hphotos-ak-ash3/t31.0-8/10012698_453226301474826_470179409_o.jpg

Thanks.

[reddeercity]

@Reddeer
You mentioned card warm up off in a previous post as you think Lexar do not benefit from this? Still stands?

Yes and no, what I mean is I see no benefit's to card warm ups to continuous record but, as the card gets Hotter
It seems to stabilizes at a higher writing speeds with less buffering. I really see it when the camera hit about 64-67 degree C
and in Crop mode when I push it at 2048x1024 and can get about 2000+ frames. But I think when I change from small jPeg photo in
the Canon menu to small raw that's when I Notice a Big difference in performance. I would say 15-20% even to be the
difference between continuous and Not.
With the Oct24th build of course

[Yoshiyuki Blade]

@romainmenke

I'll give it a shot: a1,b2,c2
They are indeed extremely difficult to tell apart and my choices are without much confidence. I didn't do any zooming though, but had to look closely. The false color has a much greater negative effect than any loss in resolution.

[JosiahDuncan]

I push it at 2048x1024 and can get about 2000+ frames. But I think when I change from small jPeg photo in
the Canon menu to small raw that's when I Notice a Big difference in performance. I would say 15-20% even to be the
difference between continuous and Not.
With the Oct24th build of course

Are the Lexar Cards really that magic or am I doing something wrong? I got a little excited and tried to test things out again but can't get anywhere near those speeds, i have to step down to 1728x972 to shoot continuous 16:9 on a Transcend 32gb 1000x. Or instead I shoot 1880x800 anamorphic and upscale to 1920x816 for 2.35. Last summer I assumed Transcend had really good build quality, but does the brand of the card make that much of a difference?
Tried on both Oct24th Build and the Newest Build

[Audionut]

You can also use the width tag

[img width=900]

[reddeercity]

Are the Lexar Cards really that magic or am I doing something wrong? I got a little excited and tried to test things out again but can't get anywhere near those speeds, i have to step down to 1728x972 to shoot continuous 16:9 on a Transcend 32gb 1000x. Or instead I shoot 1880x800 anamorphic and upscale to 1920x816 for 2.35. Last summer I assumed Transcend had really good build quality, but does the brand of the card make that much of a difference?
Tried on both Oct24th Build and the Newest Build

Yes I thinks so, Lexar 1000x cards seem to work better with 5d2, but my 64GB card is faster then my 32GB.
I always get better number then most, I usually shoot at 2:1 or 1.85:1 (1873x936 mlv & 1880x1014 raw)
Crop mode for continuous at 2048x930 (that's my favorite size)
*Note* even thou I can record at 16x9 (large frames 1880,1872) its not stable with HDMI devices, I choose to use 1.85:1 or 2:1
with those sizes . 

Edit: Also check your card for being fragmented , I notice when the card slow down a little it's always fragmented by about 30%.
If you are working in Windows , I do a disk defragment & the speed always return . FYI the maximum write speed I have seen on
my 5d2 with Lexar 1000x is 79.8 MB/s ,but only in Crop mode 1:1 is around 75 MB/s on average.
For MLV try the Nightly Build from Feb16th, that seems to be the Fastest I have found to work with 5d2 @ 1872x936 2:1 .
That's the one I use for paid work, after that date things started to slow down and became unstable.

[romainmenke]

You can also use the width tag

[img width=900]

Thx for the edit.

[romainmenke]

spoiler: http://romainmenke.com/ml/a.html
pored it into a little html page. not perfect (the scroll doesn't follow) but the only way I could let you see them on top of each other in 5 minutes work.

another confession: I scrambled the naming in both sets I posted. Since no one noticed that, I'd say there is virtually no real world difference between 1856 and 1880.

crops:
a1 oct 24 1880
a2 nightly 1856

b1 nightly 1856
b2 oct 24 1880

c1 oct 24 1880
c2 nightly 1856

full size:
a1 nightly 1856
a2 oct 24 1880

b1 nightly 1856
b2 oct 24 1880

c1 nightly 1856
c2  oct 24 1880

[xvince1]

Stating uppon 3 screens...

Aside that, I'm not a pixel peeper, but 1880 to 1856, there's a loss of 24 pixels (something about 1.3%). It's very few, I definitly agree, but there's a loss... what the point ?

As it's seems possible (and relatively easy to add), I still don't understand why not including the option ?

[romainmenke]

Like it was stated before the choice is not between 1856 an 1880.
The choice is between simple implementation for all camera's or an exception for the 5d2.
This exception comes without real coding effort now but makes it harder to maintain Raw for the 5d2 in the future.
The simple implementation has a very slight loss of resolution but makes ML easy to maintain, bug test, ....

So if there ever is a bug in raw video or a new feature there is no need to wait for the 5d2 port. It will just be there from the start.

a1ex asked for proof that 1880 is the way to go and worth all the extra trouble.
I think 1856 is just as good as 1880 even if logic says that there is loss of resolution and quality.

the loss is 1,0265 percent