Hey Tim,

Maybe this is a stupid question so forgive my ignorance. On your pre settings on the HD100U, be it Film Noir or otherwise, why not wait until post to add those touches? Don't you get stuck with a pre setting and much more difficult to change once it is shot?

In other words, doens't it make more send to shoot it perhpas more flat and then do all "film look" in post later?

Well, I'm not Tim, but I could offer my opinion on the matter. When painting your image in camera, it's good to be sure that's the look you want. The advantage of getting the look in camera as opposed to in post, is that color correction introduces image degradation. The more cc you need, the more degradation. If you have got your look down in camera, that’s less degradation to your sensitive signal. If you were working with a 4:4:4 system like the Viper, that's irrelevant. But working with HDV or even DVCPROHD and HDCAM to a lesser degree, that's an important thing to consider. Specially in HDV.

Well, I'm not Tim, but I could offer my opinion on the matter. When painting your image in camera, it's good to be sure that's the look you want. The advantage of getting the look in camera as opposed to in post, is that color correction introduces image degradation. The more cc you need, the more degradation. If you have got your look down in camera, that’s less degradation to your sensitive signal. If you were working with a 4:4:4 system like the Viper, that's irrelevant. But working with HDV or even DVCPROHD and HDCAM to a lesser degree, that's an important thing to consider. Specially in HDV.

That makes sense to me =) I knew the answer was there somewhere..

Well there are two schools of thought on the subject of when to colour grade video. I have traditionally subscribed to one, but with the advent of HDV I may be rethinking my position in some cases.

I have always been of the opinion that I should capture as much information (widest latitude/dynamic range) as possible when shooting video so that I can manipulate it any way I want in post. "Digital Negative." Some may disagree with me, but this theory works great with Viper/Dalsa (4:4:4) HDCAM (3:1:1), DVCPROHD (4:2:2) and most of the time DV and DVCAM (4:1:1 in NTSC, 4:2:0 in PAL)
I've never understood the value of sitting behind the "coffin" colour correcting the signal from a F900 while shooting on set. In my mind that doesn't save time or money. Colourists/timers are specialists, and they are paid alot to do what they do in a controlled environment. DPs shouldn't be "committing" to the look on a field monitor while on set. It just doesn't make sense to me.

I've been amazed in the past at just how much information is actually retained in the DV codec, and I've been able to bring up low to mid levels without too much added grain, colour banding, or artifacts. I feel that the DV codec is very robust considering it is only 25Mb/sec.


However since I started shooting with the HD100 and have started pushing its low-light limits, I have been very disappointed in how the MPEG2 codec handles blacks. The colour sampling is still 4:2:0 but I see much larger "steps" and colour banding in dark gradations. There is also more colour banding in the mids than I'm used to with DV. Depending on post-production workflows, extreme colour corrections can bring out the worst in the MPEG2 codec.

Now, since the camera can process the signal before the encoder encodes it, you will have much better results in some situations.
Personally, I have still consistently been shooting with my WIDE LATITUDE setting and crushing my blacks back down in post and making other minor adjustments (crushing is always fine - the problems arise when you want to raise the levels of low-mids.) However, if I was shooting for a director who knew he/she wanted to have a "Bleach Bypass" look, then I would probably just load my scene file for bleach bypass and then we would know what to expect - but it is risky because there would be no going back!

So to answer your original question, yes I generally do shoot wide dynamic range and colour correct in post.

Tim....

we ran into some quality-control issues when it came to crushed blacks when our project went through Warner Bros for dvd release. Similar portions of our film were flagged when we blew up to 35mm. Sub-level blacks were not acceptable, so I am curious if when you refer to crushing the blacks, you somehow restore the black level to legal levels at some other point in post?

I should also mention that our black level violations were very small... only a percentage or two out of range, yet we were instructed to correct it.

It would seem lowering black levels to mask noise, artifacting, or other issues.... isn't going to be an option if you plan on any sort of commerical distribution.

Networks are just as picky when it comes to spots or programming.

My question as well Tim. How far are you pushing the color and dumping the blacks in post? I find the black to be very predictable. The biggest problem I find in color correction is saturation. If you push saturation too far, the macroblocks will make your image look like a checkerboard. There is no escaping that with any mpeg.

I don't know how you define 'low' light? 20 lux? 60lux? 5 lux? Are you shooting Dogme style? Available light and reflectors only?

So I am curious if when you refer to crushing the blacks, you somehow restore the black level to legal levels at some other point in post?

I should also mention that our black level violations were very small... only a percentage or two out of range, yet we were instructed to correct it.

It would seem lowering black levels to mask noise, artifacting, or other issues.... isn't going to be an option if you plan on any sort of commerical distribution.

Networks are just as picky when it comes to spots or programming.

Everytime someone mentions crushing blacks I keep wondering WHY!!!

Now I'll admit I'm biased against the trend toward proceesing every frame. Thank god CBS execs finally forced CSI NY to cut the blue look. NYC doesn't look like that! It was silly use of CC. I'm so biased, I see CC only to fix problems, not to create a "look." I want an HD window into reality.

But, that's me. What if I did want a "look." How do I know ahead of time where this scene may fall? How do I know the exact look?

Even more to the point -- how can I trust a 3.5-inch LCD to represent the adjustment? Of course I can't! So one needs a HQ HD monitor. And, a waveform monitor.

So if I can't really adjust in the field, then I want the widest possible dynamic range. That is done by Stetching Black at 3 with the Knee at 80. And, running the highlights up to 108 IRE.

This is why I keep stressing shooting as though one was using negative film. One wants all the shadow detail one can get! (Use no gain to avoid noise that creates black blocking.) Also I want all the highlights with no bleach-out.

--------------------

The key to keeping quality in post is a very high quality intermediate codec -- like CineForm's 10-bit CFHD.

---------------------

By the way, my tests indicate the HD100 has an ASA of 200. Which means it is much like shooting Kodak Vision2 200.

I subscribe to the "Digital Negative" school of thought, except I don't think its a negative. I think its old school slide film- a "digital positive" if you will.

Anyway, I try to do as much of my grading in post as possible. I switch settings in the camera under the same circumstances I would switch film types.

HDV has caused me to reconsider this though, just like Tim. HDV is a very poor codec. You have far less latitude to "fix it in post" than you did with DV25 formats, as Tim alluded to. (I suppose that makes this a "me too!" post.)

So- unless you are willing to capture scenes over component analog I would recommend painting the image in camera.

I still recommend being conservative though. Get in the neighborhood with the camera image controls. Pick your house and redecorate in post.

Now that I've had my turn of phrase let me be clearer: I try and set up HDV to be as close to my intended final image as possible while still maintaining good technical qualities. (like exposure latitude.) When in doubt I lean towards clean technical video. This is because I find it easier to fix clean video- or to change my mind and do something different.

Finally, you have to think about your post workflow. Even though I may be shooting HDV (aka The Bad Codec), I capture footage I intend to retouch significantly in DVCPRO HD, or even uncompressed if I have the drives needed on that project. That doesn't make the original images better of course, but it gives your color grading system more latitude than if you work with HDV directly. (I color correct/grade with Final Cut 5 and Shake)

With a good workflow even The Bad Codec produces good images at output time.

As an aside, MPEG2 isn't to blame here. MPEG2 is just a wrapper format. The checkerboarding you see when codecs fail is in part an attribute of MPEG2, but it is still fundamentally the codecs failure. MPEG4 is better, but this is in great part to the coincidence that people making MPEG4 formats are also using far newer/better/more efficient codecs.

If you are stuck with a bad codec, try upconverting before color correcting. This improves blocking effects. The problem is that some of it is always going to be in the source footage- which will be faithfully reproduced of course.

Still if you are using something like Shake or Smoke for grading/correction you get the benefits of things like "optical flow" and it gets minimized- often dramatically.

Steve,

I think over manipulation of black levels (or any levels for that matter) is essentially an attempt to compensate the picture in a subjective manner.

Feature films and many TV shows are much darker today (that is to say, more areas of black on the screen) than 10 or 20 years ago. I know many producers that think they can replicate that feel by simply lowering black levels.

I would love to have a LCD waveform monitor in the field (there must be something like that available?)..... and your points about black and brightness levels in the field is supported by the fact that doing otherwise will forever commit you to a loss of detail that you may find wanting in post.

Tim....

we ran into some quality-control issues when it came to crushed blacks when our project went through Warner Bros for dvd release. Similar portions of our film were flagged when we blew up to 35mm. Sub-level blacks were not acceptable, so I am curious if when you refer to crushing the blacks, you somehow restore the black level to legal levels at some other point in post?

When I say "crush" the blacks, I don't mean to bring them below legal levels by lowering the video setup level (aka Master Black.) What I'm getting at is that when shooting with the widest latitude/dynamic range possible you may end up a very low contrast image with some "milky blacks," depending on where you chose to expose your image. Sometimes the end result is that your blackest shadows are only just "grey" but they really should be black. A simple lowering of the blacks in post (not "setup" level) will remedy this, make everything look sharper, and won't affect the 50~100 IRE range.

My question as well Tim. How far are you pushing the color and dumping the blacks in post? I find the black to be very predictable. The biggest problem I find in color correction is saturation. If you push saturation too far, the macroblocks will make your image look like a checkerboard. There is no escaping that with any mpeg.

I don't know how you define 'low' light? 20 lux? 60lux? 5 lux? Are you shooting Dogme style? Available light and reflectors only?

I seldom increase saturation. If my image requires "poppy" colours I would shoot with standard matrix and a little color gain in-camera.

I see what you're saying Tim.... it's just the term "crushed black" is synonomous with "illegal" in the experience that I have.

This discussion has given some points to ponder, because our QC required some raising of black level even though our setup was right on the money. I'd have to get a scope in front of me, but is it possible to lower blacks below the setup level? That doesn't make sense.....

Anyways... I didn't mean to suggest you were breaking "the rules"... in fact, we all seem to be saying the same thing. Sorry if I gave the wrong impression.

Everytime someone mentions crushing blacks I keep wondering WHY!!!
Artistic aesthetics. To each his own.

By the way, my tests indicate the HD100 has an ASA of 200. Which means it is much like shooting Kodak Vision2 200.
Yes. 200ASA when using standard gamma - but nothing like the latitude of Vision2 - its more like the old 200ASA EXR stock.

Actual resultant sensitivity is dependent on the gamma level setting, just like the F900/950 or Varicam. We already discussed this in another thread.

http://www.dvinfo.net/conf/showpost.php?p=381200&postcount=10

The whole thread: http://www.dvinfo.net/conf/showthread.php?t=54054

Everytime someone mentions crushing blacks I keep wondering WHY!!!

Now I'll admit I'm biased against the trend toward proceesing every frame.
I want an HD window into reality.

But, that's me. What if I did want a "look." How do I know ahead of time where this scene may fall? How do I know the exact look?

Well, if you want a look, then that's what you want.

If you have no idea where the scene your shooting fits in the end product, then you are already screwed- or doing a documentary/news.

The former means you should seriously rethink your workflow/production techniques. The latter means you have no business posting video as dramatically as we are talking about.

I think you may be in/come from more of that latter situation from the way you are reacting. Don't alter it- it represents the reality you captured. That's cool and entirely valid.

I am the opposite- moving images should be completely molded to my will. I hope you realize that is a valid viewpoint too.

Needless to say we will reach different conclusions about technique because our intentions are divergent.

Even more to the point -- how can I trust a 3.5-inch LCD to represent the adjustment? Of course I can't! So one needs a HQ HD monitor. And, a waveform monitor.

Well, you are right- you can't trust the viewfinder for just about anything except framing. So yeah bring the monitor and the waveform, and the vectorscope.

We are all a bit spoiled by using SD for so long. The viewfinder was good enough. Heck some viewfinders are better than the TV's viewers would see the final product on.

In HD that is 100% false. The best most expensive HD viewfinders will betray you utterly every time.

So, for HD, a monitor and scopes are just a standard part of the kit now.

So if I can't really adjust in the field, then I want the widest possible dynamic range. That is done by Stetching Black at 3 with the Knee at 80. And, running the highlights up to 108 IRE.

This is why I keep stressing shooting as though one was using negative film. One wants all the shadow detail one can get! (Use no gain to avoid noise that creates black blocking.) Also I want all the highlights with no bleach-out.


That's good advice in general, but it doesn't help achieve artistic effects. Not everyone is intent on creating a window into reality.

So, if a monitor and scopes ARE part of the kit then you CAN adjust in the field, and do so VERY accurately.

Now that I clicked preview I see Tim already responded nicely, so I quit here.

I will just add that despite wanting to do exactly whatever the heck I want I try to be a soft touch with any sort of color manipulation.

I see what you're saying Tim.... it's just the term "crushed black" is synonomous with "illegal" in the experience that I have.
Yes. I guess in the video world the term has referred to the sometimes used practice of adjusting the setup level of video equipment. In teleciné colour correction sessions it is simply a term used to communicate "make the shadows darker."

This discussion has given some points to ponder, because our QC required some raising of black level even though our setup was right on the money. I'd have to get a scope in front of me, but is it possible to lower blacks below the setup level? That doesn't make sense.....
It is totally possible in North American NTSC because setup is supposed to be 7.5%. However, since Japan uses 0% setup, NTSC gear needs to be able to do both. In the old broadcast days we used to have something called "super black" that was technically illegal, but a broadcast switcher could interpret it as a "alpha channel" and key over live footage. This is why is has always been important for NTSC setup to be at 7.5%.

When I scoped the HD100 I actually found it really hard to make it go into illegal black - but it was super easy with the DVX100.

The real question is "does HD always have a 0% setup - no matter what country?"

If you are stuck with a bad codec, try upconverting before color correcting. This improves blocking effects. The problem is that some of it is always going to be in the source footage- which will be faithfully reproduced of course.
Since the codec has limitations (as they all do), I've always subscribed to getting the image the way you want "in camera". The HD-100 is completely capable of getting the images the way you wish right in camera. This eliminates massive color corrections in post in favor of more suttle changes if needed in the NLE's color corrector.

I think setting up the camera to represent the final intent (as closely as possible) is the best method for digital delivery.

So- unless you are willing to capture scenes over component analog I would recommend painting the image in camera.

I'm not sure what you mean.

I can transcode to an intermediate codec (uncompressed) in the camera and send the signal via a component analog cable where it is digitized

OR

I can send HDV via FireWire and transcode in the computer to an intermediate codec.

The point is that the concerns about manipulating HDV simply have no validity if one uses a high-quality intermediate codec.

However, if you plan to use native HDV or AIC then you likely do have to worry. That means FCP users would probably be wise to use Tim's idea of a 1280x720 4:2:2 8-bit DVCPRO HD codec as the intermediate codec if you can't use 10-bit uncompressed.

Steve.

Alexander's point is that once the camera encodes the signal into HDV on tape, the damage has been done.
Yes, it is a good idea to transcode into the best codec possible (uncompressed 10-bit 4:4:4) if extensive manipulation is required and you won't be adding any more compression artifacts.

Alexander was probably referring to a live uncompressed recording device like the Wafien that will catch the signal before the Mpeg2 encoder can do its work.

I am the opposite -- moving images should be completely molded to my will. I hope you realize that is a valid viewpoint too.

I completely agree. But your concept means you really have to decide where to do the molding.

"Well, you are right- you can't trust the viewfinder for just about anything except framing. So yeah bring the monitor and the waveform, and the vectorscope."

My impression is that except for the well funded that just isn't an option. So if you can't bring all this equipment to the shoot -- do you dare adjust based on the LCD/VF? I think we both agree that could be a big risk.

Since the codec has limitations (as they all do), I've always subscribed to getting the image the way you want "in camera". The HD-100 is completely capable of getting the images the way you wish right in camera. This eliminates massive color corrections in post in favor of more suttle changes if needed in the NLE's color corrector.

I think setting up the camera to represent the final intent (as closely as possible) is the best method for digital delivery.

Generally I agree, but I don't do this by painting the image in the camera.

I try to create the image I want in front of the camera.

What I think we are talking about is when that can't be achieved.

That's where we disagree.

Most cameras don't paint as well as you can in a good NLE, much less any sort of grading solution.

The controls are simply not refined enough.

So, its sort of a middle road. I do color the image in camera, but always with an eye towards keeping my options open in post. That means technical correctness first and foremost every shot.

Oh, I am not being very precise here at all.

Steve.

Alexander's point is that once the camera encodes the signal into HDV on tape, the damage has been done.
Yes, it is a good idea to transcode into the best codec possible (uncompressed 10-bit 4:4:4) if extensive manipulation is required and you won't be adding any more compression artifacts.

Alexander was probably referring to a live uncompressed recording device like the Wafien that will catch the signal before the Mpeg2 encoder can do its work.


Thanks, I probably would have explained that in a forty page essay. Saved everyone there. <grin>

While I am a proponent of using uncompressed video everywhere possible, with HD that is rarely feasible. So, with HD I try to use DVCPRO HD. Sometimes even that is not possible and I have to go to HDV.

With uncompressed I feel confident with just leaving the image alone in camera- nice clean technically correct capture of what has been created on set. I won't color in camera unless I have a very very clear image of the end product in mind.

With DVCPRO HD I feel compelled to work a little harder on set to create exactly the look I want. I will color in camera for many shots, but I try to limit it in preference of the better controls I will have in post.

With HDV I feel the image is corrupted once The Bad Codec gets its paws on my image... so I try to get as much of what I want in the end product in camera before the codec does its brilliant technological terror thing. (Now I feel somehow compelled to speak about the power of the Force.) Better post controls are no use if your source material is decimated.

Even with the imminent threat of The Bad Codec looming I still try to keep options open for post.

You never know when that happy shot of mom looking out the kitchen window will be reused as a melancholy or pyrrhic shot showing her ready to survive some tragedy. Ahhh Kuleshov you rascal.

I completely agree. But your concept means you really have to decide where to do the molding.

"Well, you are right- you can't trust the viewfinder for just about anything except framing. So yeah bring the monitor and the waveform, and the vectorscope."

My impression is that except for the well funded that just isn't an option. So if you can't bring all this equipment to the shoot -- do you dare adjust based on the LCD/VF? I think we both agree that could be a big risk.

Well in my last installment I think I hit all that about when to mold the image. The problem is that where I feel we should do the molding slips around as the technological environment and other constraints hit our hypothetical production.

As to not being able to afford a monitor, waveform and vectorscope to the shoot- well if you can't afford these things then you can't really afford to shoot HD properly.

My feeling is that you could do away with the vectorscope and waveform on set if your monitor is properly calibrated. You need a monitor though under most circumstances.

First where don't you need it ? ENG mostly, some documentary stuff.

You can make do on those shoots with autofocus and the focus assist or peaking or whatever its called on your camera.

Any other type of work means you need a monitor. Minimum requirement is that it show the 2/3 of the full resolution of your format. That's minimum not suggested. Can't tell you how many times the XL-1(original) viewfinder tricked me- it has 215000 pixels about 2/3 of SD. BAD.

Guess what ? None of the viewfinders available for HD meet that criteria- not even on Panavision CineAlta equipment. The HDVFC30W displays 960x540, or a mere 1/4 of the total image resolution all for the economical price of $10,500 USD.

http://bssc.sel.sony.com/BroadcastandBusiness/DisplayModel?m=0&p=2&sp=19&id=69518

I have a strong preference for the largest LCD screen that is practical to tote to the set. It has always been worth the effort.

The main point is that if you can't get the camera of your choice and a monitor you can use in the field at a minimum then you can't afford to be in the HD production game.

Save your money and WAIT. Maybe better viewfinders will turn up, if nothing else new stuff will show up lowering the costs of all the "old" stuff.

Generally I agree, but I don't do this by painting the image in the camera.

I try to create the image I want in front of the camera.

What I think we are talking about is when that can't be achieved.

That's where we disagree.

Most cameras don't paint as well as you can in a good NLE, much less any sort of grading solution.

The controls are simply not refined enough.

So, its sort of a middle road. I do color the image in camera, but always with an eye towards keeping my options open in post. That means technical correctness first and foremost every shot.

Oh, I am not being very precise here at all.
This "get the color right in camera" is relatively new for me. Before cinegama came along, I was in alignment with your thoughts on the matter. When cinegama came along, it changed the way I approach color (capture). I'd rather get the colors the way I want on tape then wait for a color correction render any day of the week. It's now the way I work. At most I'll pull the blacks down to suit myself.

Of course it depends on the type of work you do as well.

I'd rather get the colors the way I want on tape then wait for a color correction render any day of the week. It's now the way I work. At most I'll pull the blacks down to suit myself.

Of course it depends on the type of work you do as well.

That is a great way to approach things, but as a practical matter, at least for a feature, you are never going to avoid the need for color correction. You can get it close in-camera (or in come cases, achieve the look and feel of the visual), but for the final matchup from scene-to-scene, even shot-to-shot, requires a scope and tweaks.

Blacks in particular can be troublesome when relying on your eyes rather than a scope. You may have black levels shooting up and down with each edit. If you are seeking mass distribution or transfer to film.... you will be doing those renders anyway.

That is a great way to approach things, but as a practical matter, at least for a feature, you are never going to avoid the need for color correction. You can get it close in-camera (or in come cases, achieve the look and feel of the visual), but for the final matchup from scene-to-scene, even shot-to-shot, requires a scope and tweaks.

Blacks in particular can be troublesome when relying on your eyes rather than a scope. You may have black levels shooting up and down with each edit. If you are seeking mass distribution or transfer to film.... you will be doing those renders anyway.

I wholeheartedly agree. Even if you intend to make it perfect in camera you won't.

This is part of why even when I am using HDV I leave at least some room for post adjustments. To me that means tending towards technical correctness over artistic sensibility on occasion.