abrupt highlights clipping

[Daniel Alexander]

Originally Posted by Michael H. Stevens

Glenn: I have read all the links but still would like some specific info related directly to the EX1 such as:

1) If I capture with Clip-Browser, what color space are the mxf files in? (To me they act like computer RGB)
2) If I set my project to 32bit how does that change the mxf files color space?
3)If I capture with Cineform I have a studio RGB, right? So how will footage look or act different from mxf files as in 1 above.
4) If I render out for Blu-ray what color space does the burner get or expect?
5) Do these question indicate I'm totally wacko?

Good questions

[Piotr Wozniacki]

Originally Posted by Bob Grant

No problem but before I do I think that using either PAL or NTSC would be a mistake, there's no such thing as PAL or NTSC in HD, in SD there is a difference between REC 624-2 for PAL and NTSC (slightly different colourspace). There is no such difference in REC 709, Vegas uses the Component profile for all HD mpeg-2 encoding. This would explain why you're seeing a difference, basically you're changing something that you shouldn't.

In fact the use of the terms PAL and NTSC in relation to HD is wrong, it makes more sense to use 50Hz / 60Hz or Region 50 / Region 60.

I would think so, too - but my experience with both HDV from the V1E and MXF from the EX1 shows otherwise. If I leave NTSC (Vegas default), I'm getting washed-out renders. It's true that other than NTSC, all other settings (i.e. PAL, Component, etc.) seem to produce similar results (i.e. with colours and black level identical to my raw clips before rendering in Vegas).

In fact, my first BD I burnt from Veags timeline seemed so much washed-out that I checked the Blu-Ray templates, and discovered they ALL had NTSC as the video format in the Advanced video tab (even the only PAL BD templates, i.e. the 1920/50i and 1440/50i ones). I changed them both to PAL, and am getting beautifully saturated BD's now (all this in 8bit projects, and without the colour space conversion to Computer RGB, which does the same for the preview window).

[Bob Grant]

Originally Posted by Piotr Wozniacki

I would think so, too - but my experience with both HDV from the V1E and MXF from the EX1 shows otherwise. If I leave NTSC (Vegas default), I'm getting washed-out renders. It's true that other than NTSC, all other settings (i.e. PAL, Component, etc.) seem to produce similar results (i.e. with colours and black level identical to my raw clips before rendering in Vegas).

In fact, my first BD I burnt from Veags timeline seemed so much washed-out that I checked the Blu-Ray templates, and discovered they ALL had NTSC as the video format in the Advanced video tab (even the only PAL BD templates, i.e. the 1920/50i and 1440/50i ones). I changed them both to PAL, and am getting beautifully saturated BD's now (all this in 8bit projects, and without the colour space conversion to Computer RGB, which does the same for the preview window).

Sorry but all the BD templates I just checked show Video Format "Unspecified" and the Color primaries etc as Rec 709. Which is I believe exactly as it should be.

Saying that your first disk looked washed out is a pretty meaningless statement really. That could be due to a number of factors or combination of all of them. Without knowing where in the process this is happening and adjusting something could lead you down the road to all manner if problems. I know the old saying "All that matters is how it looks on the big screen" and if that's all that you're after then tweak away. Just be aware it might look very different on another big screen.

Simple suggestion. Record bars in the camera. Go through your whole process using the bog standard settings for everything and then calibrate your monitor(s) and HDTV from those bars.
Better still if you can get a BD calibration disk and calibrate your HDTV if you haven't already done so. That's got to be your first step.

[Piotr Wozniacki]

Bob,

I checked the 8.0b original templates and you're right - "Unspecified" is this field value. But when the BD burning from timeline first appeared (i.e. in the original VP 8.0 release), it was NTSC - I remember for sure as it was then when I run accross the colour saturation problems for the first time (it coincided of course with the introduction of 32/2.22 F.P. processing, and made the things really complicated).

Now, it might have been an SCS bug; indeed I have just finished a quick render with video format "Unspecified" rather than "PAL", and it looks OK (i.e. same as PAL, more saturated than with NTSC).

If this is the case, it looks like I'm back to a more comfortable situation I used to be in with Vegas 7, when I didn't have any doubts regarding the colour space whatsoever. Meaning that - unless a project involves heavy edits that actually involve colour re-calculations - I can follow exactly my workflow used with HDV and Vegas 7 (stay in 8 bit, do not use RGB conversion, have anything BUT NTSC in the final render's template Advanced Video tab).

Should need arise, I'd be switching to 32bit mode and follow Glenn's colourspace conversions table... If this all works as expected, than the only doubt left would be the Vegas preview window colours, and Vegas scopes, credibility...

Looks like a plan to test out thoroughly now!

[Piotr Wozniacki]

Just as a follow-up:

I have turned the colour management on, and checked the Studio RGB on the Windows Secondary Display properties sheet - what a difference! When I click the "external monitor" icon above Vegas preview window, before the monitor switches into the overlay mode, I can see it increase the colours' saturation and dynamic range - what I can see full-screen is now exactly the same punchy colours and levels as in the raw MXF's. All this _WITHOUT_ unnecessary converting the colour space in my mxf's, and in 8bit mode...

I definitely recommned doing it to all complaining about the Vegas preview window showing washed-out colours, before fooling around with colour space RGB conversion, or engaging the time consumming 32-bit processing.

Of course, if someone uses a separate monitor, it will do the trick as well. Thanks Glenn!

[Bill Ravens]

Originally Posted by Piotr Wozniacki

Just as a follow-up:



I definitely recommned doing it to all complaining about the Vegas preview window showing washed-out colours, before fooling around with colour space RGB conversion, or engaging the time consumming 32-bit processing.

duh!

sorry, Piotr, I couldn't resist ;o)

[Glenn Chan]

1) If I capture with Clip-Browser, what color space are the mxf files in? (To me they act like computer RGB)
2) If I set my project to 32bit how does that change the mxf files color space?

8-bit Vegas project: studio RGB
32-bit Vegas project: computer RGB

3)If I capture with Cineform I have a studio RGB, right? So how will footage look or act different from mxf files as in 1 above.

Decodes to studio RGB, regardless of what mode your Vegas project is in.

4) If I render out for Blu-ray what color space does the burner get or expect?

It expects Y'CbCr values. You need to make sure the Y'CbCr values are encoded properly. So you need to know what type of RGB-->Y'CbCr conversion is being done. Or stated differently, the encoder will expect either studio or computer RGB levels. You need to feed it those levels.

One way to achieve that is to nest your .veg, and then apply a color space conversion if necessary (sometimes it is not necessary). No color space conversion is necessary in your case, if you're not using Cineform (and assuming you ONLY have EX1 footage in your timeline; if you bring in some other formats, they will need color space conversion).
In an 8-bit project, you should make sure fades to black are fading to 16 16 16 RGB black (not the default 0 0 0 RGB). Lay a solid color generator on the bottom-most track.

5) Do these question indicate I'm totally wacko?

Nope :)

[Piotr Wozniacki]

Originally Posted by Piotr Wozniacki

Well, so I've set up some basic PPs for the most typical shooting scenarios (PP1- PP4 using the Hisat matrix with slightly differing settings):

1. PP1, based on the STD1 gamma (for indoors, low light/low and dull scenery)
2. PP2, based on the STD4 gamma (for indoors, low light but contrasty scenery)
3. PP3, based on the CINE1 gamma (for outdoors, high light but flat scenery)
4. PP4, based on the CINE4 gamma (for outdoors, high light, contrasty scenery).
5. PP5 (for outdoors low light flat; TBD)
6. PP6 (for outdoors low light backlit/contrasty; TBD)

Just an update:

Today I tested my above mentioned "Relelease Candidate" PP6 preset for outdoors at dusk; semi-dark, no additional lighting, sky 100% overcast with those dark gray clouds being the brightest part of the scenery. What I did was basically use the PP4 (see above), but further stretch blacks (Black Gamma up at 40), and compensate for that with increasing overall colour saturation (Hisat Matrix with level as high as 75). Frankly, I was afraid what I shoot would be watchable at all; it certainly wouldn't with my old good V1E.

Imagine my nice surprise when I got quite watchable, saturated video, with trees clearly against the sky without the "abrupt clipping" (thank to the Cine4's low and gradual knee), and the dark, dull foreground stretched enough to show all details, yet without virtually any noise (I managed to keep gain at -3dB for the first 15 mins, than had to go up to 3dB as it was getting dark fast). I started with the shutter of 1/50th but, since it was a low-light handling experiment, I sacrificed some movement fluidity for the extra stop, and when it became even darker switched the shutter off (1080/25p with 25fps).

Of course the grab below is certainly not the best picture that could possibly be obtained, but I think it's a step to the right direction. This camera can really "brighten the reality"...

[Piotr Wozniacki]

I decided to put it in this thread, as many others - also dealing with various aspects of the EX1 picture settings - have been heavily biased towards their interactions with what Vegas does, and how it displays what it does :)

OK, so I must say that - after having tested all gamma curves in various conditions - the ugly patches of colour, resulting from how the camera handles light in the vicinity of the knee ("abrupt highlights clipping") can actually happen with ANY gamma curve.

Of course, one can avoid this by carefully watching the picture and zebra/histogram and either decrease or increase exposure to avoid the tricky area. This is now always possible, however.

Even though the Cine curves (especially, CINE2) are more forgiving than the Standard ones, they have fixed Knee settings which prevents the user from tweaking them to his preferrences.

I wonder, since nobody is reporting similar observations, do most people just accept it, or is my camera (for whatever reason) more sensitive to the phenomenon?

Also, is it just me, or does every unit brighten the picture for split second (like some 2 frames) when switching between auto iris on and off? I find this annoying, as - when I switch the iris to Auto - before it gently adopts the new aperture value, the exposure goes up for no apparent reason (sometimes by as much a couple of stops - it's only visible in the picture, though; the aperture reading in the LCD doesn't reflect it).

[Paul Kellett]

?

Also, is it just me, or does every unit brighten the picture for split second (like some 2 frames) when switching between auto iris on and off? I find this annoying, as - when I switch the iris to Auto - before it gently adopts the new aperture value, the exposure goes up for no apparent reason (sometimes by as much a couple of stops - it's only visible in the picture, though; the aperture reading in the LCD doesn't reflect it).[/QUOTE]


Bill,mine does that aswell.
Paul.

[Alexander Ibrahim]

Originally Posted by Piotr Wozniacki

...

OK, so I must say that - after having tested all gamma curves in various conditions - the ugly patches of colour, resulting from how the camera handles light in the vicinity of the knee ("abrupt highlights clipping") can actually happen with ANY gamma curve.

Of course, one can avoid this by carefully watching the picture and zebra/histogram and either decrease or increase exposure to avoid the tricky area. This is now always possible, however.

Funny I remember reading this sort of statement elsewhere.

If it isn't possible to set exposure properly, then the next step is to modify the lighting, or the cameras view of that light- again as I've said here before.

Some combination of netting, scrims, neutral density and other tricks to bring that portion of the image under control.

If you are shooting something without the budget and/or time to implement those lighting controls then your clients, and the audience, aren't likely to notice or care.

I wonder, since nobody is reporting similar observations, do most people just accept it, or is my camera (for whatever reason) more sensitive to the phenomenon?

This happens with ANY camera... from your eye to a fisher price pixelvision, and yes even the EX1.

Its pretty hard to notice with your eye though because your eye is equipped with the best auto iris ever- oh and don't forget a ridiculous dynamic range... but you can see it.

So... do I notice it? Sure, in most every camera I've used in a similar situation. Do I complain? Nope. I either correct the exposure or modify the lighting. If I can't do any of that then the shot wasn't worth worrying about and I accept it.

Also, is it just me, or does every unit brighten the picture for split second (like some 2 frames) when switching between auto iris on and off? I find this annoying, as - when I switch the iris to Auto - before it gently adopts the new aperture value, the exposure goes up for no apparent reason (sometimes by as much a couple of stops - it's only visible in the picture, though; the aperture reading in the LCD doesn't reflect it).

I hadn't noticed that... but then I haven't looked. The first thing I do with an EX1 is turn off auto anything. I don't turn them back on.

Leaving aside my peculiarities, why on earth would you change from manual to auto iris during a take?

[Piotr Wozniacki]

Alexander;

I cannot agree with you this time:)

Regarding the "abrupt clipping" - it's untrue that "every camera does that". My V1E didn't. The only thing to worry about when filming a back-lit scene was choosing what I wanted to expose properly: the bright backgroud so it doesn't clip, or perhaps a person standing against it so it's not under-exposed? That was quite normal; the decision was mine. With the EX1, a third factor is involved: how to avoid the clipped colour patches I can't even see in the LCD while shooting? I certainly consider the way it handles it as being flawed; it's not just the question of inevitable clipping! For instance, can you tell me why are those patches always bluish - regardles whether it's clipping the blue sky, or a gray cloud?

Also, with regards to auto iris on/off thing: very often, when following my subject in an uncontrolled scenery, I'd just switch to auto and let the camera decide while I'm busy with framing and focussing. Then, having re-framed and re-focussed, I'd like to take the exposure control back from the camera.... Again, with the V1, all I had to do was push a button; the camera would take / release control gracefully. But not the EX1 - it will open the iris momentarily, severely overexposing 1-2 frames, and only then set the appropriate aperture.

So frankly I really DO consider these to be flaws in practical use outside a controlled environment like a studio. And one more thing: you said elsewhere that PP's (and the tweakable parameters in general) should serve artistic purposes, not "fighting" for a usable picture... Well, I wish you were right!!! I also would be very glad, but it simply isn't true. Without tweaking, the picture from my EX1 is much worse than that from the V1E I had (also tweakable, but in much more straighforward and predictable fashion). The only obvious advantages over my previous camera are then reduced to better low-light sensitivity and better MPEG-2 codec - but NOT colours and latitude! This is less than what I was expecting, therefore I'm trying to find best PP for each type of lighting conditions, in order to optimize the EX1 picture. Only when (or if?) I succeed, will I play further with the picture parameters for the "artistic purposes"!

There is only one possible explanation, or if you like: excuse, for the above "peculiarities": if the EX1 is treated as a stationary camera for shooting under carefully controlled light, and with enough time to set-up each take - the flaws (including sticking ND switch) become irrelevant. With ENG use, however, they are really serious (is there anything you could call "a take" in run'n'gun shooting?). With its handycam form factor, has the EX1 been designed as a stationary-only (or -mainly) camera? From what you are advising, Alexander, I take it that you're advocating to use it that way...

PS. Oh, and Alexander - even if you never switch auto iris on during shots, could you just check it for me, please? Perhaps it's just my unit - you don't need to actually shoot and play the recording back; it's enough to observe the iris ring: when I slide the Iris switch into the Auto position, the ring does a very jerky movement, and only then slowly and gently adopts the new position (the second, proper movement's range depending of course on the difference betweeen the current manual setting, and what the camera thinks it should be).

Same request to everyone else... TIA!

[Bob Grant]

Yes, my EX1's iris does a jump when I switch from manual to auto. Thinking it through I might be asking for my money back if it didn't. This was advertised as a camera with a full manual lens. That means the operator can control the lens through purely mechanical means i.e. no electronics between human and lens. Actually I think we got dudded with the zoom but I guess given the huge cost of real zooms I can live with that.
Anyway, for the iris control when switched to manual the servo motor has to be powered down or mechanically disengaged. The latter is what happens from memory on the big broadcast lenses. In the EX1 I think the servo motor is simply power off, you'd sure never move the iris ring if it was powered up.
When power comes back on (you switch to Auto) the the motor will jump, something about coils of wire and magnets come to mind from high school physics.
The V1 is a different beast from memory, not a real manual iris. In manual you takeover from the feedback loop in the exposure servo control circuit, the iris servo is still powered up. When you switch to auto the control goes back to the feedback control from the exposure metering circuit.

You wanted a better camera than the V1, you got it. Does not mean it'll be easier to use, that was a given as far as I was concerned. Take the next step up in cameras and you'll find there's no autofocus. Next step or two up the rung and you'll find auto iris doesn't exist either.

[Piotr Wozniacki]

Thanks Bob for confirming the iris thing. Yes - the first thing that came to my mind when I first noticed it was: this is the price I have to pay for the fully manual iris ring...

Please do not get me wrong; I'm not bashing the camera. I'm simply trying to point out that opinions such as Alexander's (that it behaves exactly like any other camera would) are simply wrong: it does not! And those people who - like myself - are asking questions in order to understand it better, are fully entitled to do so, rather than treat everything specific to this camera as "normal".

I had some doubts and problems with my V1E as well, if you remember - so I hope that after some more time I'll make friends with the EX1 just like I did with its predecessor. If I regret anything, it is only that I had to sell the V1E as keeping both would not have been "economically justified" in my case - the two would complement each other perfectly!

[Piotr Wozniacki]

OK, so - following the conclusions in previous posts - I'm asking those more knowledgeable to help me in the following:

Having established that the CINE curves are more forgiving than the STANDARD ones, I'll keep my PP's build around the CINE gammas as the safe settings. However, in case I'd like the punchy look that only STD1 offers, while trying to avoid "abrupt clipping patches" -

- would you move the KNEE point up or down?
- would you make the curve above it more or less sloppy?
- would you increase or decrease the compressed highlights' saturation?

Any opinions welcome!