Copeing with Out of Range Scopes

MH_Stevens wrote on 2/16/2008, 8:42 AM
With cameras that record super blacks and whites (like the EX1) the Vegas scopes seem to be redundant as they do not show the signal above or below the 0-100 scale and it is not possible to see where either the top or bottom clips. (Piort/Megabit has screen grabs over on the DVi EX1 forum if you don't know what I mean - "Abrupt Highlight Clipping" post #186)

Mike

Comments

GlennChan wrote on 2/16/2008, 12:32 PM
My guess is that this behaviour occurs because the values are being converted to 8-bit, and then processed by the scopes code.

Perhaps something you can do is to take the levels filter and put values into range... this will let you see if clipping is occuring or not. Or if there is extra detail to be gained.

2- Perhaps you can put in a feature request for the scopes to be re-designed so that they do show values that are out of range.

(Personally I believe that Vegas should do some sort of automatic colorspace management. Currently, you have to consult a table like the one on my website and that's just complicated!

I think it all kind of boils down to that solution... Vegas handling colorspace issues for you, similar to what FCP does. But none of the stupid gamma issues... which is unique to Quicktime.)
MH_Stevens wrote on 2/16/2008, 6:43 PM
These out of range values are not white clipped. No clipping in the image. Do not others have this problem?

Very confused.
farss wrote on 2/16/2008, 7:03 PM
"Do not others have this problem?"

I've seen it and no I don't see it as a problem.
Set your scopes to Computer RGB, Vgeas displays the full range of possible values. Same image in Photoshop reads the same values.

It's possible that how Vegas is decoding to R'G'B' is wrong but that's another issue entirely. If it is then it's not alone.

Bob.
GlennChan wrote on 2/16/2008, 9:36 PM
These out of range values are not white clipped. No clipping in the image.
The scopes are clipping values when they shouldn't right? As in, what's displayed on the scopes is clipped; the underlying values are still there and can be brought into legal range.

Or stated differently... suppose you send the signal to real scopes and compared that to Vegas' scopes (and assume that Vegas' scopes are setup correctly). Vegas' scopes would give the wrong reading in this case.
MH_Stevens wrote on 2/16/2008, 9:43 PM
So the answer to my question is "you can't cope with Vegas out of range scopes.' Seemes very strange to me that now I can only gauge clipping in Vegas by eye and the scopes are redundant.

Why are they wrong?

GlennChan wrote on 2/16/2008, 10:44 PM
It's because the scopes are receiving 8-bit R'G'B' values where white level is at 255 and black level is at 0 (0-255 range).

This is a smaller color space than 8-bit R'G'B' with a 16-235 range.

Which in turn is a smaller color space than 8-bit Y'CbCr with a 16-235 range for Y'.

There is clipping happening when the color space conversions are performed. It's going from 8-bit Y'CbCr to 8-bit R'G'B' (0-255 range) which causes the scopes to clip.

Seemes very strange to me that now I can only gauge clipping in Vegas by eye
You're dealing with superwhites.

Usually they will get clipped, but not always. CRTs can display them, and not all equipment will try to intentionally clip them off.

You can bring them into legal range in Vegas, in which case they won't get clipped off later on in the delivery chain.
farss wrote on 2/16/2008, 11:59 PM
Glenn,
I think what's being suggested elsewhere is that the EX1's Y'CbCr is outside the 16-235 range for Y'. If it's going to Y' = 255 then wouldn't Vegas be clipping it during the conversion to R'G'B' ?

That's going to be hard digital clipping and not recoverable.

Also what I'm confused about by what Michael is saying is regardless of what mode the Vegas scopes are in they don't show clipping unless what Vegas has decoded is clipped. In Studio they might show values over 100 but they're still not clipped, as in flatlined. In that case yes, they can be bought back into legal range and nothing down the track will clip the signal. I don't think this is the concern. The concern is that the camera is not saying the video is clipped and yet when it's decoded by Vegas it is.

For the record I'm not seeing any of this at all but there's so many ways to tweak the image in the EX1 that probably doesn't mean much at all.

Bob.
craftech wrote on 2/17/2008, 7:03 AM
I kind of gave up on the scopes a few years ago and now do all color correcting by eye. Is that wrong? So far I haven't noticed a problem and I have tried my DVDs on lots of different televisions both CRT and LCD.

Of course, I don't do broadcast work.

John
MH_Stevens wrote on 2/17/2008, 9:03 AM
DO understand this. None of my images are clipped in reality. I have perfect detail even in the hottest whites. No areas of white blocking when the Vegas scopes show major clipping. I have asked Piotr to tell me how to take a screen shot of the scopes and then I will publish (at DVi as I can not do it here) the scopes and the picture. I may also take an image of the EX1's histogram and then that should show all.
GlennChan wrote on 2/17/2008, 10:35 AM
I think what's being suggested elsewhere is that the EX1's Y'CbCr is outside the 16-235 range for Y'.
Most cameras do that.

If it's going to Y' = 255 then wouldn't Vegas be clipping it during the conversion to R'G'B' ?
Not really.

In both 8-bit and 32-bit Vegas projects, you can map the superwhites back into legal range (e.g. use the Levels filter).

If you don't do that, then the superwhites are liable to get clipped. But they kind of are supposed to get clipped.

2---Vegas' scopes can be wrong. I think that is what the original poster is seeing. The scopes are wrong in that situation.
Bill Ravens wrote on 2/17/2008, 11:14 AM
I REALLY hope you can get these guys straighted out, Glenn. Between here and DVINFO, MHStevens and Megabit are taking up a LOT of bandwidth. These guys are like a bad dog with a pants cuff. They shake it until something falls out.
farss wrote on 2/17/2008, 12:18 PM
It's understandable. If you look at the whole picture from what the camera is doing, what its metering systems are saying, through to what Vegas is saying and doing. Throw into the mix the question of semantics and you have the ingredients for a lot of misunderstanding.
Personally the biggest challenge I find with the camera is my own abilities. Trying to get focus right is a challenge.

Bob.
megabit wrote on 2/17/2008, 1:20 PM
Why so patronizing, Bill?
If all myself and Michael are asking about is so obvious to you, why wait for Glenn to "get these guys straightened out"?

AMD TR 2990WX CPU | MSI X399 CARBON AC | 64GB RAM@XMP2933  | 2x RTX 2080Ti GPU | 4x 3TB WD Black RAID0 media drive | 3x 1TB NVMe RAID0 cache drive | SSD SATA system drive | AX1600i PSU | Decklink 12G Extreme | Samsung UHD reference monitor (calibrated)

Bill Ravens wrote on 2/17/2008, 5:14 PM
Patronizing? c'mon. you guys are given the answers to your q's, and you don't see what you're being told. You both need to spend some time with your cameras shooting, editting, rather than trying to think your way thru this stuff, and getting everyone to explain things. Believe me, it WILL come to you. It's really not as hard as you're trying to make it be.The WFM is nice, but, look at the histogram in Vegas. It WILL tell you if you're out of bounds for the DISPLAY. The EX1 histogram is only a guide for the CAMERA exposure, NOT for eventual display. Without RGB channels in the histogram, it's not accurate. Are your monitors calibrated? If they're not, you can't tell anything relative to final display. I'd believe the scopes before I'd believe the preview window in Vegas. The "right" answer isn't single valued. It's multivalued. That's why it's called art. Beauty is in the eye of the beholder....yaddah, yaddah.Sometimes zebra is the right "guage", sometimes the light meter is, sometimes the histogram. It really depends on the characteristics of the scene you're shooting. Experience will tell you what method to use with what scene. It take time and practice. That's partly why people spend a LOT of money on film school.

If videography was "paint by the numbers" holy mackerel, we have no hope to be better than the next guy. Some of the images I saw you guys post looked damn good. The WFM distracted you from the nice images? It's easy to get lost in the forest of the minutiae.
Geez, I bore myself, sometimes.
GlennChan wrote on 2/17/2008, 7:11 PM
megabit/piotr:
You have some legitimate questions... please don't be afraid to ask them!

Sometimes Vegas does some unintuitive stuff and sometimes some discussions can get confusing because it's hard to figure out what context/semantics other people are using.
Serena wrote on 2/17/2008, 7:43 PM
Just back in the office and really should switch on Vegas to check before speaking (but have to go out again) but my Vegas scopes show below zero and above 100 (110). Obviously I must have done something incorrectly, but it works fine for me. Yes, the EX runs up to 109 if I have exposed to that and that's what I see on the scopes.
MH_Stevens wrote on 2/17/2008, 10:37 PM
Serena: You have done nothing wrong (like you could) - this is just the quandary that has got a lot of us very confused and asking with HD and 32bit in Vegas, just what is clipping. It has become a philosophical question.
farss wrote on 2/17/2008, 11:47 PM
Oh, we're talking 32bit!!!!

See how we loose sight of the trees for the forest.

Well yeah, for sure. I have a clip that's deliberately seriously overexposed. In 8 bit the Vegas scopes read 95ish (in Computer) but STILL some signs of detail. Switch to 32 bit and it reads 100% and I mean bang up against the stops flatlined. Wow that looks so clipped it's fried.

However add the Levels FX set to ComputerRGB to StudioRGB to the preview window and the scopes now read like they did in 8 bit and the image in the preview window looks like it did in 8 bit. I take no credit for this discovery. I just read what Glenn was saying, all of it, carefully. I tested, tested, thought about things until it made sense.

Are the Vegas scopes broken? I'm not so certain, how should they display the difference between 0.999997 and 0.99998 ?

And what does in mean in practice, that's what really matters. Reality seems to be that most LCDs clip the tops and the bottoms of the gamma curve. It might be nice for us that there's still detail there but don't count on the viewer seeing it. Knowing how to deliver great looking images onto the big screen and the little screen takes more than just peering at scopes. As Bill said it takes a friggin lot of experience. If it didn't even I could do it.

Bob.
Serena wrote on 2/17/2008, 11:52 PM
>>>It has become a philosophical question.<<<

Ah, I see. Matters of the mind.
GlennChan wrote on 2/18/2008, 7:21 AM
It has become a philosophical question.
I don't think so. There's a difference between Vegas' scopes and hardware scopes. If you need to do broadcast work, then the difference is important. But then again, I'd just get a set of real hardware scopes and not trust Vegas'.

Vegas' scopes will clip values when it shouldn't.

2- You also have to set them up correctly and you have to interpret them correctly.

If your the codec you plan on rendering out expects computer RGB levels, then in the Scopes settings, make sure the studio RGB setting is unchecked.

The "7.5 IRE" setting should correspond to whether or not your country uses analog setup (or the country you're delivering to uses setup). However, if you turn this on, there is no marking at 7.5. So this makes it very difficult to judge black level. It's also difficult to tell whether or not this setting is on (do you see how this might be dangerous?).

Vegas' scopes can be useful as an aid in CC... I prefer to set it up incorrectly (leave the 7.5 IRE setting unchecked, even though I live in a country that uses setup)... then I can use the scopes to help judge black levels when adjusting levels.
MH_Stevens wrote on 2/18/2008, 9:52 AM
Serena:
1) I say it is a philosophical question because this thread challenges the definition of just what clipping is. Having scopes that flat-line way short of clipping, what now is the definition of clipping?

Bob, Glenn:

2) Changing from Computer to Studio RGB only drops the black level on the scopes. The top end (in my footage) stays the same, still showing clipping.

3) Changing the scopes settings, 7.5IRE or Studio check box, (for my EX1 footage) DOES NOT change my scope readings or in any way compensate for a change in color space.

4) If I use Sony Levels to bring the scopes into range the right and left sides are still showing a sudden drop-off - no gradual bell curve diminution as is normal. So what does doing this show?

5) If I change to an 8-bit project then the width of the scopes is narrower but that in it's self only further proves the scopes useless. Changing project settings can not change the reality of how footage is exposed. So once again I ask, what value the Vegas scopes now and just how is clipping defined. (and I'm talking science here not art. I have no problem getting good pictures)

Mike

GlennChan wrote on 2/18/2008, 11:19 AM
2) Changing from Computer to Studio RGB only drops the black level on the scopes. The top end (in my footage) stays the same, still showing clipping.
Changing from Computer to Studio RGB may not be appropriate. The purpose of that is to get color space conversions right (which may not be called for).

Superwhites (the illegal values that most cameras generate/shoot) are a separate issue.

2- In a 32-bit project (NOT 8-bit), applying a levels filter with output end = 0.907 will bring superwhite detail into legal range. In a sense, that will avoid clipping (though this depends somewhat on what context you are using).

3) Changing the scopes settings, 7.5IRE or Studio check box, (for my EX1 footage) DOES NOT change my scope readings or in any way compensate for a change in color space.
Ack, I wasn't clear. It will affect the Waveform display, but not the histogram. This is what I see... you should be seeing the same thing.

4) If I use Sony Levels to bring the scopes into range the right and left sides are still showing a sudden drop-off - no gradual bell curve diminution as is normal. So what does doing this show?
I'm not sure what you mean there.

5) If I change to an 8-bit project then the width of the scopes is narrower but that in it's self only further proves the scopes useless.
I'd say that the Vegas scopes are sometimes not helpful.
MH_Stevens wrote on 2/18/2008, 1:02 PM
****************MAJOR UPDATE**************

Thanks Glenn: After more testing I have found that I can get the Vegas scopes to agree with the EX1 histogram if I use:
** An 8-bit project
** And apply the Computer to studio RGB conversion.

But now I get the washing out that Piotr reported, and to correct it needs the good old "S" curve in CC.

So, forget the theory this time, what settings should I be using in Vegas? Do I forget the scopes and use 32-bit and computer RGB, Do I go 8-bit and use computer RGB or do I go Studio RGB and then have to color correct every image from the camera which are all wonderful under computer RGB?

And if I render to Blu-ray disk how will these choice effect the render. IE under what circumstance is the Vegas preview screen close to WYSIWYG?
farss wrote on 2/18/2008, 1:56 PM
As far as I'm aware there's no 7.5 IRE setup in NTSC digital video. The A>D converters will remove it if setup correctly and the D>A converters will add it if setup correctly. The only reason I can see why the Vegas waveform has it there is some converters possibly don't work correctly or you may set them up incorrectly and therefore end up with digital video with setup.

Changing the Vegas waveform scope between computer and studio only changes the scales not the waveform.

Connecting a hardware waveform monitor to the composite output of the EX1 shows no superwhites. If I can get hold of one it'll be interesting to see what a HD waveform monitor shows connected to the HD-SDI output of the camera.

Keep in mind that Vegas works in R'G'B'. All values per channel from 0 to 255 are 'legal' including out of gamut colours. What may happen to them down the path is another question. Vegas's waveform scopes are showing you the Y' value when the R'G'B' are converted to Y'CbCr. As far as I know not every output codec makes the conversion the same way and different display devices handle the values differently and maybe calibrated (if at all) differently.

In the most basic strictest sense digital clipping occurs when there's no more digital values left. If you have a signal that goes from 0 to 1.0V and you have an 8 bit A>D converter that is setup to give 0 to 255 over that range and you apply a signal over 1 volt it's clipped, 1.0 volts and 1.10 volts both give a value of 255. Now we might all agree that we'll call 0.90 volts 100% of something, let's say engine thrust, that's rough 230. Someone tweaks the engine a bit and gets more thrust giving a value of 240, nothing has clipped but an analogue meter connected to a D>A converter that's scaled to 230 = 100% might have a hard stop, we'll not see the extra thrust on that meter.

Bob.