50D Raw video

[dogmydog]

Andy please give a hint what is it? My impatience is exploding my mind I really need to know that my buying the second 50D and shooting with them both commercial stuff is cool )))))))))

I read about it on twitter.
https://twitter.com/autoexec_bin

Andy, please don't be mean and tell us! I cant sleep lolol

[djfremen]

Wonder if it's for TL only? lol

[lionelp]

I would rather not speculate.  But the ML tweet indicated the ML release was something to do with the sensor and Andy indicated that his end of the surprise release was something to do with post.

[painya]

New Codec for the 50d perhaps?

[djfremen]

Maybe they figured a way to encoded directly into a QuickTime container. //end speculation.

[Andy600]

The ML developments will hopefully gain a little more DR from the sensor but not sure yet if this is coming to the 50D.

My company's product is post-production related and works across a range of cameras, not purely Canon, but ML raw video really benefits from using it. I can't say much more at this stage because it's going out to reviewers over the next week.

@dogmydog - sleep? what is this 'sleep' that you speak of? 

[painya]

I saw that on the "multiples of ISO 160 gaining more DR" or something like that. Everybody should go check that out. Warning, it's pretty techy.

[Audionut]

But my curve shows the connection between light and the reading from the sensor. We all work with the reading from the sensor in post-production, not with the light itself.

You curve shows the curve that your post processing software, or Canon picture style is applying to the linear raw data.

The raw data is always linear.  The connection between light and the reading from the sensor is linear.  Since human eyes are not linear, a tone curve is applied to make the output more closely resemble what human eyes, see.

[rommex]

You curve shows the curve that your post processing software, or Canon picture style is applying to the linear raw data.

The raw data is always linear.  The connection between light and the reading from the sensor is linear.  Since human eyes are not linear, a tone curve is applied to make the output more closely resemble what human eyes, see.

So, from your words, the raw data contains not lighting measurement itself, but log(2) thereof. That's clear.

But if RAW data is only about EV, why such curves from Canon and BMCC (imported in the same way to the same software) differ. Why these sensors have different DR? If RAW data had only linear EV data, all sensors would behave the same way.

[Audionut]

Why these sensors have different DR? If RAW data had only linear EV data, all sensors would behave the same way.

All sensors behave differently because they have different pixel well capacities, electronic noise, linear gain, etc, etc. 

http://mikeboers.com/blog/2013/11/07/linear-raw-conversions

[Andy600]

All sensors behave differently because they have different pixel well capacities, electronic noise, linear gain, etc, etc. 

http://mikeboers.com/blog/2013/11/07/linear-raw-conversions

Noise, sensitivity, calibration and the intended use of the sensor data dictates the curves that are applied by manufacturers and software devs for sometimes not purely technical purposes. Strip away the curves so the readout is linear and they all basically function the same but with +/- sensitivity and noise depending on the sensor.

[rommex]

All sensors behave differently because they have different pixel well capacities, electronic noise, linear gain, etc, etc. 

http://mikeboers.com/blog/2013/11/07/linear-raw-conversions

Thanks!

While being linear in the main section, raw values do differ in the foot and in the top (judging by dcraw graph). And while some curve is applied in software, these differences just change the form, but they stay.

This is exactly the pivotal point of my curve -- how it ends up and down. And my curve is exactly the thing that thousands of people operate with in their post-production.

So thanks for clarification.

[rommex]

.... but not sure yet if this is coming to the 50D.

[Andy600]

I haven't looked at the registers but hopefully 1% will get on it

The 50D would probably benefit most from this upgrade because it has the lowest DR of the raw video shooting cameras.

[Andy600]

Re: my earlier tease - here's a frame grab from the 50D (crop) with only a hint or NR, a curve and a little sharpening added. Very minimal work to get this. Nice detail?



click for full size.

[rommex]

Re: my earlier tease - here's a frame grab from the 50D (crop) with only a hint or NR, a curve and a little sharpening added. Very minimal work to get this. Nice detail?

wow... was it crop mode? what ISO?

[dogmydog]

The ML developments will hopefully gain a little more DR from the sensor but not sure yet if this is coming to the 50D.

My company's product is post-production related and works across a range of cameras, not purely Canon, but ML raw video really benefits from using it. I can't say much more at this stage because it's going out to reviewers over the next week.

@dogmydog - sleep? what is this 'sleep' that you speak of? 

wow... was it crop mode? what ISO?

Looks like crop mode to me.
+1 on the ISO question

[rommex]

This particular area is particularly interesting. It's hell of a dark, it's 50D, and we have such gradations of tonality here...

[Andy600]

Yes crop mode (1920x872). ISO 400 on a 28mm at F4. I should have shot at a much higher ISO but the shadow detail is still pretty good.

[lionelp]

From my small screen (my phone) I don't see signs of noise.

[dogmydog]

I'm about to shoot a clip on a beach in the next days.
My worries is the excess of light and DR. Hence some questions:

1) Should I use a low ISO like 100 or try to beam it up to get more DR?
2) Can we get higher shutter speeds then 1/48 when shooting raw or MLV raw? I don't have ND* filters, only a B+W Polarizer I intend to use.
3) Camera heat... I'll use a 7D as a backup, with no magic lantern, shooting H264 if raw on 50D fails. I'm thinking about bringing some ice packs on a box to cool it off when not in use.

Maybe thats a little off topic, but it's 50D and raw video use, so any tips will be highly appreciated

[1%]

You can use the 7D with raw too. The res is only slightly smaller in 1x. The raw still looks good even if upscaled 2x, unlike the H264.

I think with FPS the shutter speeds go to 1/110 or something like that in 24p

[rommex]

I'm about to shoot a clip on a beach in the next days.
My worries is the excess of light and DR. Hence some questions:

1) Should I use a low ISO like 100 or try to beam it up to get more DR?
2) Can we get higher shutter speeds then 1/48 when shooting raw or MLV raw? I don't have ND* filters, only a B+W Polarizer I intend to use.
3) Camera heat... I'll use a 7D as a backup, with no magic lantern, shooting H264 if raw on 50D fails. I'm thinking about bringing some ice packs on a box to cool it off when not in use.

Maybe thats a little off topic, but it's 50D and raw video use, so any tips will be highly appreciated

Yep, usually lower ISO secures the widest DR available for the sensor.
In my 50D I can set up 1/147s as the shortest shutter speed.
I shot a commercial with 7D H264 in summer 2 years ago. Overheating was huge (

[rommex]

PLEASE HELP!

It's late and I'm slow to find a solution:

I have the latest TL and I used to have this bw playback of raw files. But now I can't find it. Stupid me ((( Tomorrow morning is the shoot....

[Andy600]

PLEASE HELP!

It's late and I'm slow to find a solution:

I have the latest TL and I used to have this bw playback of raw files. But now I can't find it. Stupid me ((( Tomorrow morning is the shoot....

Press the play button (top scroll to scroll through takes). Controls for playback are accessed with the PicStyle button. B&W + color is selectable for raw and mlv files.