5D2 RAW video Builds 14-Bit

[ch_d]

I´ve got one question on my 5D2 and the Komputerbay 1000x 64GB card.
Here is my benchmark:

Since i got about 75MB/s writing speeds on 1880x1058 (and even in crop mode about 77MB/s) is my card good enough even with this benchmarks? ... since the camera speed limit is reached.
or should i replace this card?

[Africashot]

... since the camera speed limit is reached.
or should i replace this card?

Why would you replace it?

[Doyle4]

You will also find when recording your write speed increases, this is because its warming up.. iv done a benchmark before and write was 67mb, when it came to recording it hit 71mb.. test it on RAW video first before sending it back, remember, the benchmark is only 5mins long.. so you wont see the cards full potential.

[ch_d]

Why would you replace it?

because the benchmark results are too slow ... ? aren`t faster cards faster in terms of transfering to the computer / copying files?
i saw way better results on the 64gb 1000x komputerbay card

[Doyle4]

because the benchmark results are too slow ... ?
i saw way better results on the 64gb 1000x komputerbay card

record a 4gb+ video, doesn't matter what but keep on eye on write speed, what do you get?

Also some devs have said to format the card before recording,

[dariSSight]

Have anyone read anything on timecode ability for ML, also I think I read we can set our own presets now so can we keyword soon?

[mvejerslev]

Why are you all benchmarking in photo mode?

[ch_d]

Why are you all benchmarking in photo mode?

here is one in video mode

[Doyle4]

Benchmark in Playback:

turn on camera, then press info till screen is off (black) then run benchmark, should say Playback, not video or photo

[ch_d]

Benchmark in Playback:

turn on camera, then press info till screen is off (black) then run benchmark, should say Playback, not video or photo

so here is one in playback mode:

[Africashot]

because the benchmark results are too slow ... ?

I had not realized you bench marked in photo mode, this way the card indeed looks slow, I assume you must have tried formatting too... was it any better at any earlier stage, do you get better results with other cards?

[Doyle4]

so here is one in playback mode:

Much better and thats good speeds for the markii

[melihozbek]

Hi,

With the new build, bolt_rec.mo gives an error and won't load... I checked it from bitbucket, it changed 4 days ago, but I failed to find the compiled version. There is only bolt_rec.c .. does anybody point me to the latest compiled version of bolt_rec.mo? I tried to compile but, it exceeds my abilities..

Thanks,

Melih

[a.d.]

it works well with ce9fb256205c.zip
BOLT_REC.MO

[melihozbek]

Thank you a.d.

[ValentinSteiner]

What ISO are you guys using when shooting 5D2 Raw?

A couple of weeks ago I did a quickt test to see how the Raw video changes depending on the ISO setting and I found that the higher the ISO, the less noise was visible. Can that be right?

If I remember correctly, I also saw a kind of "grouping" of certain ISOs; meaning that the actual exposure was similar between multiple ISOs.

[reddeercity]

What ISO are you guys using when shooting 5D2 Raw?

A couple of weeks ago I did a quickt test to see how the Raw video changes depending on the ISO setting and I found that the higher the ISO, the less noise was visible. Can that be right?

If I remember correctly, I also saw a kind of "grouping" of certain ISOs; meaning that the actual exposure was similar between multiple ISOs.

I find the the higher you go with ISO the more chance of noise,
But as long as you don't under expose the image use what ISO looks good to you
i use negative (-0.3) or push down ML digital ISO.
So at 400 ISO (canon) becomes Magic Lantern digital 320 ISO .
There is little to no noise with push down ISO's
These are may experiences with ML ISO's
Other people may have some more advice to add.      

[robert.roth001]

Well the push down ISOs aren't actually magiclantern ISOs, they're completely canon's doing. I saw in a thread (something posted about a month or so ago) that the digital ISOs are exactly the same in noise performance to their native ISO, when shooting raw. Using the pull down ISOs definitely work better in H.264, but there is no benefit in terms of noise performance in raw.

What I mean by this is that 160, 200, and 250 are all the same noise performance in raw, with 160 being .80 exposure of 200 and 250 being 1.25 of 200. Not 100% sure if that's how you would make the analogy, but that's the basic idea. So 320, 400, and 500 are all the same as well, in raw. It just keeps going on. This is how I understand it at least. I haven't done any tests with it yet, but that's what makes sense with understanding the hardware and looking at other people's tests.

[Kharak]

What ISO are you guys using when shooting 5D2 Raw?

A couple of weeks ago I did a quickt test to see how the Raw video changes depending on the ISO setting and I found that the higher the ISO, the less noise was visible. Can that be right?

If I remember correctly, I also saw a kind of "grouping" of certain ISOs; meaning that the actual exposure was similar between multiple ISOs.

This "grouping" you speak of, I might be able to testify to that.

Cause a week ago or so me and a friend of mine were comparing extreme low light performance between my 5dii and his FS100.

We had ISO's ranging from 12000, 6400 and 3200. But when I looked at the DNG's the extreme ISO scenes were all at the same exposure.
6400 and 12000 are pushed from 3200 native iso. they were all exposed like the 3200 ISO. But in camera they looked brighter. Quite weird, but I guess there isn't too much difference between pushing ISO in camera and pushing it in ACR. I've had great results in ACR.


PS. I beat his FS1000. Some of his scenes seemed brighter and vice versa, but the 5D RAW II had much sharper images. FS100 records in AVCHD (standard) with some kind of "box" it can record AVI Uncompressed 12 bit, not sure how that compares to RAW.

[reddeercity]

Well the push down ISOs aren't actually magiclantern ISOs, they're completely canon's doing. I saw in a thread (something posted about a month or so ago) that the digital ISOs are exactly the same in noise performance to their native ISO, when shooting raw. Using the pull down ISOs definitely work better in H.264, but there is no benefit in terms of noise performance in raw.

What I mean by this is that 160, 200, and 250 are all the same noise performance in raw, with 160 being .80 exposure of 200 and 250 being 1.25 of 200. Not 100% sure if that's how you would make the analogy, but that's the basic idea. So 320, 400, and 500 are all the same as well, in raw. It just keeps going on. This is how I understand it at least. I haven't done any tests with it yet, but that's what makes sense with understanding the hardware and looking at other people's tests.

Well form my video samples i  tested i have found that if you use negative gain the image  is very clean No noise.
Being most of the noise is in the Luma channel, The black & white of the image.
So when you push the expose in Post i find there is more latitude in the shadows & darks.
From my understanding when you use the ML digital gain ISO's there are a different algorithm , i think no sure though.
The results of this is that i do not have to use any noise reduction software at all, no matter what iso i use.  

[PressureFM]

Well form my video samples i  tested i have found that if you use negative gain the image  is very clean No noise.
Being most of the noise is in the Luma channel, The black & white of the image.
So when you push the expose in Post i find there is more latitude in the shadows & darks.
From my understanding when you use the ML digital gain ISO's there are a different algorithm , i think no sure though.
The results of this is that i do not have to use any noise reduction software at all, no matter what iso i use.  

Give us some examples, I have a hard time seeing how you could pull this off if you are already at the cameras limit, at say, ISO1600 (which is the max native ISO for the 5D Mark II) and are still underexposed.

Isn't this merely true to your style of shooting because you have plenty of available light?

In RAW it only makes sense using ISO100, 200, 400, 800 and 1600. The rest are just manipulated afterwards, e.g. ISO160 is just ISO200.

To test it, shoot something at "ML" ISO160 and  Canon ISO200, then pull down the ISO200 to 160 in post. Do you see a difference?

[Doyle4]

100,200,400 are digital iso's,

320,640,1600 are not 1600 gives less noise than 100 if done correct

[Africashot]

100,200,400 are digital iso's,

320,640,1600 are not 1600 gives less noise than 100 if done correct

Can you expand on this theory? I'd really like to know more about how to make 1600 look as clean as 100!

[SpcCb]

100,200,400 are digital iso's,

320,640,1600 are not 1600 gives less noise than 100 if done correct

I think there's a mistake somewhere

[Steadycamer]

Hi there,

Here's a test with ml raw, hope you like

https://www.youtube.com/watch?feature=player_embedded&v=8_7UKTDmr0Y

Cheers,
Vincent