XDCAM EX Picture profile settings / HDSDI questions

[midnightblue]

Hi I am a new owner of a Sony PDW-EX3, and I have a few questions after reading some of this forum and the operation manual, and I would appreciate if some of the more experienced users would take a few minutes and help out with these concerns:



1) (Cinema) Gamma settings/custom Picture profiles, do ANY of these influence how the image is actually recorded, or all they are just in-camera post effects, that could be done with better quality/flexibility in Post while editing?

2) Detail, Matrix, Knee, Gain - in camera ? could be better done it post as well?

3) The operation manual states that the camcorder is able to record the 60P at only 720P, not 1080P? Does that still apply if an Flash XDR or Nanoflash is used and video is captured via HDSDI? Would that still be 720P ?? I was just thinking that maybe the 720P limitation, was there because of the 35Mbps bandwith for the SxS cards, that would be lifted in the case of HDSDI?

4) White/Black balance? is it better to do on Location, or is it the same thing fixing it in post?



Generally my goal would be to acquire 100% of what the camera can record, without any destructive in-camera effects, and would prefer to do all the grading in Post. And if someone could recommend a 100% 'neutral/pure' profile setting(s)? that would be appreciated!

Thanks!

[cuervo]

Speaking very generally, all these settings influence the signal recorded.
Video camera sensors do not have the latitude that film has. Since many scenes, e.g. outside shoots, have a dynamic contrast that exceed the capability of the camera sensor, the cinematographer is given the opportunity to customize the camera sensor to read the scene in order to make the best use of the bandwidth available. That's what all these settings are designed to do. As you might expect, not a trivial process, and one that is best served by experience. Certainly, some adjustment in post can be performed, but, post can't recover data that was never recorded in the first place.

[midnightblue]

So what you are saying that ALL these settings are actually 'analog' changes, that influence how the CMOS sensor reacts to light, and is not the same if i go into an NLE or AE and apply filters/curves etc ?

Also would someone be kind enough and write a list of 'in-camera post-effect' and 'real' settings, like for example I am quite positive that Noise reductions in-camera are all 'post' and could be much better done once editing ... maybe im wrong! any ideas?

[PerroneFord]

These are not "post-effects" any more than putting 400 speed film into a camera instead of 100 speed is an "effect". I think where you are going is that you want the RAW sensor data. Currently, that is only available (from the factory) on RED and maybe a few other high end cameras. Even on cameras like the Panavision Genesis when you make the decision to record in panalog format, it's a REAL change to the recorded image and not an effect.

The EX1/EX3 do no "noise reduction" that can be influenced by the operator. The best you can do is to shoot with -3 gain for the cleanest signal possible. And it is VERY clean.

If you take care of the basics like proper exposure, white balance, snd scene contrast, you'll come home with an image you can be proud of, and one you can move around in post.

[midnightblue]

Yes, that is quite close to what i was thinking, but it is still a bit unclear that what is actually effecting the CMOS, and what is applied afterwards to the recorded data.

The RED camera RAW is a good example, if you can access the RAW data before the picture profiles applied etc, then that would mean that no picture profile is actually affecting how the sensor does anything, it would suggest that all settings are applied AFTERWARDS the data is recorded ??? which would suggest that the gamma curves in the EX series do the same?

Which would in turn mean that its best to keep a 'conservative' neutral approach, and do most of these changes in post???


Anyone have any ideas about the HD-SDI 720P vs 1080P ???

[PerroneFord]

The RED camera RAW is a good example, if you can access the RAW data before the picture profiles applied etc, then that would mean that no picture profile is actually affecting how the sensor does anything, it would suggest that all settings are applied AFTERWARDS the data is recorded ??? which would suggest that the gamma curves in the EX series do the same?

Which would in turn mean that its best to keep a 'conservative' neutral approach, and do most of these changes in post???


Anyone have any ideas about the HD-SDI 720P vs 1080P ???

Unfortunately, the RED Raw is unique. The EX1/EX3 do not work this way. Nor do any other common video cameras I am familiar with.

You are just not going to get what you're looking for. So you next best bet is to select settings that preserve as much of the image as possible. This is recording "flat" and though it doesn't make the nicest picture out of the camera, it certainly does give you a lot of room in post. I shot this way with my DVX most of the time.

I have tried to get something similar out of the EX1, and it can be done, but I have not taken the time to do it. No one I've asked has either as most people seem intent on getting the look in camera if possible.

[midnightblue]

Yes you are spot on! Thats exactly what I am looking for: settings for the 'flat out of the box' but for the most potential/flexibility in Post!

If anyone has any ideas/presets that result in that I am sure a lot of us would appreciate it!

[basspig]

Video camera sensors do not have the latitude that film has. Since many scenes, e.g. outside shoots, have a dynamic contrast that exceed the capability of the camera sensor, the cinematographer is given the opportunity to customize the camera sensor to read the scene in order to make the best use of the bandwidth available.

Actually, replace "camera sensors" with "video format" and you'd be more correct. The sensor has about 500% more dynamic range than the XDCam CODEC can record, so the reason you tweak these settings is to find the combination of knee point and gamma curves that fit the greatest amount of informaton *for a given scene situation* into the recorded format.

[midnightblue]

Thats great information basspig! So i was right for a certain extent, that it IS actually an in-camera POST-effect to an extent then that it wont change how the sensor is behaving it only affects what happens to the recorded material afterwards.

One of the members recommended using the `Standard Picture` profile, to obtain the most for Post , however after a second look at the manual this is what i found:

CINE2: This gives almost the same results as CINE1. Select this when you wish to obtain 100% video signals for editing or other purposes.

So this would suggest that this is 'best' setting?

Also one would assume that if you record through HDSDI, that is 10bit instead of 8bit, then you get more of that 500% of what the CMOS can actually 'see' ???

[basspig]

Yes, it is an in-camera post process, but to clarify about HDSDI, I would expect that signal to already have the picture profile processing applied, so you would not be seeing "raw" data from the sensor in any of these circumstances. HD SDI does not relieve the operator of the need to optimize the gamma for the shooting circumstances.

What the manual means by "100% video signals" is that the upper highlights are limited to 100IRE broadcast-legal signals. It is definately NOT what you want if maximum dynamic range is your capture intent.

What I would do is set up a profile that stretches the blacks a bit, starts and early knee on the whites and has an early slope. Then push the blacks back down in post. At least this way you're assured of capturing as much information as possible, and then having the option to adjust in post.

[midnightblue]

Excellent advice basspig, however could you explain it a bit further what you meant?

You are saying Blacks should be captured quite 'light', however the little I know of digital video signals is that dark areas can take a lot of punishment, as they record details even when they get really dark, and you will still have information in there where as bright areas get clipped sooner. That looks like it is a bit in contradiction of what you are saying but maybe i got my wires crossed and thats what you have meant ???

[basspig]

What I meant, in the topic of capturing the maximum amount of information, is that the goal is to capture a relatively low-contrast image by pushing open the shadow areas and gradually compressing the highlights. The idea is to pack as much of the original signal into the recorded video, not to optimize the "punch" of the video signal for broadcast--that would be done in post.

This could best be accomplished in 10-bit SDI signals, but I think a reasonable job of it can be done with the 8-bit files recorded to SxS.

Setting the Black Gamma for a positive value (to be determined by recording some high dynamic range scene, or maybe a chip chart) to bring the shadows up enough so that they are represented by a greater number of bits than if they were recorded normally, almost crushed, and using one of the STD gamma curves where you can set the knee point and slope, and using fairly aggressive curves here to capture a wider range of discernable highlights and pack them intact, into SxS files.

In other words, don't crush black levels and don't blow out highlights. To do this, you will give up contrast, which is what makes a picture normally have "punch" and clarity to it. It will look somewhat like the raw image off a drum scanner, before processing (I come out of a color prepress background, so excuse my references to Scitex and other hardware examples), which looks washed out and pretty dull until you normalize or optimize the distribution of luminance signals for the final delivery format.

[rodwell66]

Do you have any suggested picture profile settings that reflect your line of thought below?

Thanks!

Rodwell66

What I meant, in the topic of capturing the maximum amount of information, is that the goal is to capture a relatively low-contrast image by pushing open the shadow areas and gradually compressing the highlights. The idea is to pack as much of the original signal into the recorded video, not to optimize the "punch" of the video signal for broadcast--that would be done in post.

This could best be accomplished in 10-bit SDI signals, but I think a reasonable job of it can be done with the 8-bit files recorded to SxS.

Setting the Black Gamma for a positive value (to be determined by recording some high dynamic range scene, or maybe a chip chart) to bring the shadows up enough so that they are represented by a greater number of bits than if they were recorded normally, almost crushed, and using one of the STD gamma curves where you can set the knee point and slope, and using fairly aggressive curves here to capture a wider range of discernable highlights and pack them intact, into SxS files.

In other words, don't crush black levels and don't blow out highlights. To do this, you will give up contrast, which is what makes a picture normally have "punch" and clarity to it. It will look somewhat like the raw image off a drum scanner, before processing (I come out of a color prepress background, so excuse my references to Scitex and other hardware examples), which looks washed out and pretty dull until you normalize or optimize the distribution of luminance signals for the final delivery format.

[midnightblue]

rodwell66: I am actually looking for the same thing, the closest thing I have found is Bill Ravens' "True Color profile 2". What do you think Basspig?

[rodwell66]

midnightblue:

What are the "Bill Raven's True Color profile 2" settings you mention? I've searched the forum and can't seem to find them. Can you point me in the right direction?

Thanks,

rodwell66

rodwell66: I am actually looking for the same thing, the closest thing I have found is Bill Ravens' "True Color profile 2". What do you think Basspig?