Uh no, that isn't it. Next?

has this gone on long enough? The responses have covered the gamut of the possible. Of course a "quickie Poll" will always have a big spread so if good statistics of judgements are wanted then it will have to go on for a much longer time. Noting that "one response has come very close to the answer" perhaps there is indeed a puzzle we are supposed to solve, but that wasn't in the brief.

Serena,
There are now three very, very close answers, and I have privately promised those friends and colleagues that I will allow them to unravel the mystery, or at least start giving clues, once interest in this thread starts to wane (as would any good producer).

Are you staying with your initial statement that #2 is best and #3 is worst as viewed by you on your system and on your terms? That is all I am really interested in.

Assuming you have registered your opinions, you are more than welcome to throw the four images up on the Vegas scopes, and draw your own conclusions as to the purpose of this inquiry. At that point, it should become apparent to you as a professional, why I did not pose this as a puzzle, or even as an objective exercise, but as an open-ended poll.

But in the meantime kindly indulge me by allowing more responses to accrue before revealing your conclusions.

Until then,

OK, as you like. A sort of cryptic puzzle where we can see all the parameters but have to work out the particular question.

Yes, as you say, but with the question itself being deceptively simple.
Thanks for your patience.

Would vote for Pic #1 if it would not be your reference, it's simply the best.

Pic #2 looks overall good but dark tones are raised too much.

Pic #3 would be the best but seems to crush whites a little, in other words Pic #3 has most ooh and aah, but personally I still prefer #2. The chin in #3 looks a little saturated, and saturated faces looks amateurish in my opinion. I can live with a saturated sky (quite normal in videos due to the limited dynamic range), but not a face.

Pic #4 is worst since it seems washed out in both ends.

These are my conclusions juts by looking at the images in the browser, on a calibrated 24 inch Dell DVI-connected computer monitor.

Christian

My guess, without reference to any scopes...

Image 1 is nicely exposed, being delivered with the blacks near 0 and the highlights somewhere between 235 and 255.

Image 2 maps level 0 from image 1 to 16, but leaves 255 at 255. Hence the blacks become greyer (nothing between 0 and 16) and the highlights remain relatively unchanged.

Image 3 expands the levels both ways by mapping 16 to 0 and 235 to 255, and so ends up with some clipping at both ends.

Image 4 squeezes the levels by mapping 0 to 16 and 255 to 235, so ends up with nothing outside of 16-235 and is therefore "wishywashy".

The purpose of the test is to attempt to ascertain whether it's best to make blanket recommendations for levels manipulation that will work "correctly" for most users but will leave some users with "incorrectly" clipped levels or with "incorrectly" squeezed levels. In other words, to determine which is the lesser of various evils.

Nick,
the problem really is that the reference image is wrong. It was not taken with a video camera. Anyone who uses one of those cameras should know they either:

a) Should use a medium or low contrast setting to avoid every NLE (except for Vegas) from clipping.

b) Unlock something like Preserve RGB levels.


The rest of it well, yes, now that you've let the cat out of the bag (see my post in the other thread).

Bob.

Oh dear, the levels are taking over the asylum. We now have to top 4 threads on the forum all on the same subject :)

Don't know if I'm letting any cats out of any bags. I was just making a guess at what musicvid's up to. I might be way off the mark. Looking forward to the dénouement.

Since the wording of this video makes the answer so subjective, I' ll ask a slightly different question:

Which sample is closest to what you would expect to see if you rip any movie from your DVD library? To that question, I would answer #2. This is important to me because I want my material to look as good as possible on a system set up for playback of everything else.

On my laptop, #1 looks the best, but if I play a movie from Netflix or Hulu, it looks a little washed: like #2. If I play that same content on my Roku or Apple TV, it looks perfect. That is why I always aim for the look I see in sample #2.

Put another way:

Video should look like #1 on a tv. Unfortunately, video that looks like #1 on a TV will look like #2 on a PC. If I aim for #2 on my PC, it will look like #1 on a TV. Having an eye for getting look #2 on your PC will let you achieve look #1 on a TV pretty consistently without having to use a calibrated monitor system ((which basically just stretches the levels in look #2 into the levels you see in look #1).

Another way to achieve look number 2 is to put an extra sRGB to cRGB color correction filter on your master video buss and leave it on as you clor correct, but bypass it when you render. I actually use a a variation of this for my preview only color correction: I start off with an sRGB to cRGB color correction filter, leave the blacks as they are (stretching from 16 to 0) but I don't stretch the whites and I move the center exposure point back to the center. What this does is darken the blacks, leave the highlights alone, and keeps the center where it is. I find this to be the ideal preview only color correction and better than a straight sRGB to cRGB correction for previewing. While this gives you an accurate preview, and while I used to do this all the time, I usually just eyeball look #2 now because that extra color correction filter is just too much extra work for my aging Core2duo laptop.

Quite frankly, I am amazed at the variation in the answers. To my eyes, the answer is incredibly obvious. #2 is what we should all be aiming for in the Vegas preview window. #1 is what our video should look like on a TV when we've done this. I am surprised that everyone doesn't see it this way.

Where is Glen Chan? His posts over the years and his webpage on color is what opened my eyes on how to see it this way. Before that I was always fighting with color.

Without wishing to comment on your deductions, what you see then as a good image will depend on how well your monitor is calibrated. And should you have a "phobia" about grey blacks, you might be careful before assuming that any grey should be black. Perhaps one might also consider variables like saturation and curves.

Here is another perspective. I have been mixing sound for years now and have a reputation everywhere I work for having really good ears, this in spite of the fact that I have terrible tinnitus and high frequency hearing loss. When I mix audio, I go for a certain sound that is very much like the audio equivalent of picture sample #2. Withngood reference monitors it is very easy to come up with a sound that is wonderful in the studio but terrible everywhere else. You don't want this. You want your mix to sound good in a car, on a three inch mono radio speaker, on a large PA, on an average stereo, on a high end stereo, etc. There is an art to getting to happen.

Something with color. You want your color to be right on TV, on YouTube, on your smartphone, on DVD, on Bluray, on a projector, etc. Look #2 is the one that best does this.

It's not just how my monitor is calibrated. It's how everyone else's monitor is calibrated. It's how everyone else's TVs are calibrated. It's how a tv is calibrated when you pull it out of the box from Best Buy.

One last thing, I am typing this on an iPad2. On the iPad, # 2 looks the best. #1 looks too contrasty. #3 looks horrible and #4 is passable. One of the reasons I like the iPad so much is that it makes sRGB levels video look perfect. If I play back a YouTube or Vimeo video, the levels still look perfect even though they are cRGB and would look more like #1 on my PC. On a MacBook or iMac you would see the same thing. All the more reason to go gor straight sRGB levels like you see in #2.

"Which sample is closest to what you would expect to see if you rip any movie from your DVD library? ... Unfortunately, video that looks like #1 on a TV will look like #2 on a PC"Every time I think I understand this stuff, someone makes a statement like the Laurence's quote above that challenges my understanding. My first reaction was... bulldungy, a DVD played on my PC will look identical to a DVD played on my TV/DVD player.

Well, I was wrong. There's nothing like a test to confirm the facts. I have a Vegas Project set up for the Nick Hope Levels test:



We all know based on numerous threads that h.264/mp4 playback on YouTube or other MediaPlayers results in expanded levels such that the above video shows white on the left and black on the right.

Well I rendered the test to an mpeg2 DVD and found that while playing the DVD on my laptop, the levels were expanded - just like the mp4 playback (tested on WMP, Cyberlink, VLC). However, when I plopped the DVD in my standalone DVD player (a Samsung BDP1600), the levels where not expanded (i.e. I saw levels of white on the left and levels of gray on the right).

I love this forum 'cuz I learn something new every day.

...Jerry

Jerry,
what Laurence is saying is correct. What YT and Vimeo is doing is correct.
What PC based DVD players and VLC will do depends on how you setup your video cards drivers for "video". Vegas will let you correct computer monitors when you use them as the Secondary Display Device, Vegas "expands" the levels so the video appears correct.

The reality is in video land nothing should exist between 0 to 16 and 235 to 255, those values are illegal. Feed legal video to YT or Vimeo and it displays it correctly by expanding the levels.

If you want Vegas's internal preview monitor to display video correctly you have to expand the levels as Laurence has pointed out, as Glenn Chan has pointed out.

If you shoot video with a DSC oo DSLR your camera will almost certainly not record correct video levels and on a properly calibrated monitor it will clip the blacks and highlights. Dig into some of the vDSLR fora to read about the amount of grief that can cause.

Bob.

It's not as easy as it looks to keep your video levels legal. Drop a jpeg on the timeline without a cRGB to sRGB correction and you will be generating illegal 0 - 255 levels. Add a title with the default white and you have an illegal 255 white. That one really bothers me. Why the heck is the default text an illegal white!? Fade to black without a layer of generated 16 black on the bottom layer and you have an illegal black of 0. Easy to work around, but why do we have to? I know, I know, because we might be working in cRGB and Vegas is format agnostic... it still isn't right.

"the problem really is that the reference image is wrong. It was not taken with a video camera. "

Actually, the original image was lifted directly from correctly exposed native GH2 .mts footage and was chosen because of its wide dynamic range, high-key skin tones, and lighting range from cool open sky at 11 o'clock to the warm softbox (I assume) at 2 o'clock. By "correctly exposed" I mean it occupied a usable range of about 16-236 iirc, with the only whiteouts on the subject being a small patch on the forehead and the very tip of the nose. That is #4, which you correctly point out is not the reference print.

EDIT CORRECTION Looking again at the GH2 sample I downloaded, I see that it is Canopus MPEG-2, not native AVCHD. So I can't be precisely sure if it was leveled in-camera or in-post. However, looking at the natural dropoffs at the upper and lower ends, I would guess it is pretty much tit-for-tat from the original.

In fact, here is a histogram of the original frame before any tweaks were applied:


For this test, the dynamic range for the reference was expanded to 0-255 in order to simulate how the native "correctly exposed" footage (#4) would play back without intevention -- correctly on an RGB monitor. It also simulates the native output of many (most?) DSLRs, Point-and-Shoots, Pocket HD, Phones, hybrids, etc.

The others readily fall into place following Nick's perfect interpretation. #2 and #4 simulate 16-255 and 16-235 respectively (two common exposure variants). #3 is how #1 will play back without intervention (clipped). Only thing I left out is how #2 would play back without intervention. It is so close to #3 it's hard to tell the difference.

The discussion that has ensued in the past twelve hours is not off-topic and I encourage it to continue without injecting my own views, and I still welcome first reactions to the images, even though the bag is now empty and the cat is roaming about freely.

By asking which "mistake" people prefer without setting any ground rules, I was trying to get a feel for what people actually expect and will tolerate from their uploads. Very interesting perspectives and thought-provoking.

I will give my own impressions of the results, in a week or so. Thanks for making this so much fun, and feel free to carry on the discussion.

John Meyer, Laurence, John Dennis, Nick Hope, and Bob either got it right away or came very close, even though their choices did not necessarily agree. Imagine the six of us sitting in a bar near closing time . . .

To me this discussion is very "real world". I had so many experiences where I would do a video and it would be projected on a screen at an event and it would look horrible: all dark and no detail in the shadows, or I'd see it on somebody's TV and be shocked, or I would hate the way it looked on Youtube, etc. Thanks to what I learned from others here (especially Glen Chan on this particular subject) I now feel like I pretty much nail it every time. When I look at other videos that people post (here as well as other places), the most common mistake I see is messing up the color range. You just see people get it wrong all the time, especially now that DSLRs are all the rage.

Laurence, a lot of my interest in this came from you, going back to our discussions of what people manage to do with HDV levels.

"the most common mistake I see is messing up the color range. You just see people get it wrong all the time, especially now that DSLRs are all the rage. "

You mean like this? It was lifted straight from YT. A dangerous kid, probably with a DSLR and FCS.

"Why the heck is the default text an illegal white!? "

Yes, Vegas gets a lot of things wrong, a well known fact that we all have to work with.

The reason is probably historic. When Vegas was first written by Sonic Foundry they wer a company heavily into streaming video. This was in the days well before YouTube and Vimea, heck most of the planet was on dial up.
Back then to get video to correctly display on computer monitors you had to do a Studio RGB to Computer RGB coversion yourself.


"Drop a jpeg on the timeline without a cRGB to sRGB correction and you will be generating illegal 0 - 255 levels."

Yes, the graphic arts and digital still world is different to the video world. Graphics arts people who don't understand this are the most common cause of programs being rejected by broadcasters.

Bob.

>>>It's how everyone else's monitor is calibrated<<<
Here I think we have a different meaning for the word "calibrated". What I mean is set up according to the appropriate standard. "Adjusted" and "Calibrated" are not synonymous. While we must understand the difference between sRGB and cRGB, making allowances for individual screen adjustments isn't possible.

Image 2 is the best image but I liked image 1 the most because the subject had more feeling and intensity not too dark the picture had more depth she is in a forest so you expect it to have less midtones.
So although levels in 2 are better the picture has no depth,

the person looks happier in image 1 for some reason

Just to add
the pink in her lips is the same as her top more dominant in image 3 and 1 therefore her smile is more noticeable. that is my weird perspective so I always go with what looks good rather than calibrated levels.

some more add
And your eye stay focused on her face in 1 and 3 whereas in 2 and 4 your eyes wonder off the face and get lost in the murky well set levels