Film Processing Houses for Film Out

[David Parks]

My apologies to Tim Dashwood for not searching the older posts for film out. The question just popped into my head as I was asking about film processing.

I plan on shooting a test in the next 2 weeks (using an unbiased local DP named Gary Watson who has been shooting everything from film to HD CAM to 3/4 Umatic since the mid 70's) Great guy. I'll ask him to do a detailed write up of his thoughts and I'll share our findings.

Thanks for the detailed info guys,

[Steve Mullen]

Originally Posted by Tim Dashwood

However, I've had a conversation or two with Andrew Young at DuArt about this who also found that the HDV codec does not really maintain the video information in the bottom half of the curve the way you would hope, and the results could potentially be disastrous. The information falls to pieces and you get alot of banding and stepping.

I've covered the monitor LUT and conversion issues in my HD100 Handbook, so that is not new to me. For example, Stephen's comment "... as I said then, Tim's "Warm" scene file looked almost identicle to what I saw on my timeline" may reflect that with FILMOUT the video should NOT look correct in the Timeline. It can only look correct on a monitor that supports such a curve. In other words, if the recommendation against FILMOUT is because it doesn't look right, then that recommendation is incorrect.

I am most interested in it's origin and the lack of follow-up by JVC after the HD100 was released. JVC could have published a White Paper on it, for example.

When you say "... Andrew Young ... ALSO found that the HDV codec does not really maintain the video information in the bottom half of the curve ..." are you saying you have run your own conversion to film tests?

I'm not saying the HDV codec may not yield "... information [that] falls to pieces ..." but I'd like a lot more data than `Andrew says' before I totally rejected FILMOUT. Your phrase "the results COULD POTENTIALLY be disastrous" doesn't help much either.

Can you publish, as you have in the past, evidence that would allow filmmakers to reject FILMOUT with confidence.

I'd also to hear from someone who has used the setting used with the Panasonic monitor that has what MAY be the correct LUT.

[Andrew Young]

Originally Posted by Steve Mullen

When you say "... Andrew Young ... ALSO found that the HDV codec does not really maintain the video information in the bottom half of the curve ..." are you saying you have run your own conversion to film tests?

I'm not saying the HDV codec may not yield "... information [that] falls to pieces ..." but I'd like a lot more data than `Andrew says' before I totally rejected FILMOUT. Your phrase "the results COULD POTENTIALLY be disastrous" doesn't help much either.

Hi Steve,

I have done some testing on this and my apologies for not getting the results written up and posted. I've been traveling a lot lately. Anyway, my recommendation against Filmout is based on an actual test out to film. The results were unusable, either for post color correction or when shot straight out to film. This is not simply an issue of monitoring. The distribution of tonal information is very skewed and it appears that a lot of information gets lost in the 8 bit environment. This is not to say that Filmout might not have some utility for uncompressed analogue capture, but that still needs to be tested. Anyway, I'll try to do a formal posting on this after NAB. What sort of data would you like to see to be convinced?

[Andrew Young]

Originally Posted by David Parks

Does anyone know some good film labs that process M2T, QT, or DPX files to 35mm film for a test??

Hi David,
Sorry for not getting back to you sooner. We can work with all of those formats at DuArt, and we have done a few HD100 filmouts. Shoot me a email if you like (ayoung@duart.com). I'd be happy to talk to you about your upcoming test, regardless of where you decide to send it.

[Brian Drysdale]

Originally Posted by Andrew Young

Hi David,
Sorry for not getting back to you sooner. We can work with all of those formats at DuArt, and we have done a few HD100 filmouts. Shoot me a email if you like (ayoung@duart.com). I'd be happy to talk to you about your upcoming test, regardless of where you decide to send it.

Andrew,

I've been shooting some camera tests that we'll be running a 35mm test film out on in the near future. From your experience, could you say which camera settings seem to work best for a film out?

Brian

[Andrew Young]

Originally Posted by Brian Drysdale

Andrew,

I've been shooting some camera tests that we'll be running a 35mm test film out on in the near future. From your experience, could you say which camera settings seem to work best for a film out?

Brian

Hi Brian,

I'm in the process of trying to evaluate different menu settings for filmout. Unfortunately, I've got a lot of other things going on, so it's been a slow process. But with the help of others on the forum, we can surly get it figured out. The only basic recommendations I can make at this point are:

1) turn the detail setting down low, I recommend MIN or OFF. This is very important. The default detail setting is way too high for the big screen; 2) do not use the camera's Filmout gamma setting if you are shooting in a compressed format (tape or firewire). It is not useful in an 8-bit environment. 3) Do not use black compress. If you want a crushed black look, do it in post. 4) turn motion smoothing off and be sure white clip is set to 108%. Feel free to experiment with black stretch and knee. If you like the look of Cinelike, use it in the color matrix only, not in the gamma.

A lot of stuff I shot and filmed out was done at the camera defaults and it looked pretty good, except for the detail. The question is, how much better can it get? That’s what needs to be tested.

If DuArt does a test for you we'll deduct the cost from your final filmout. Advice is always free - that is if I have any. Good luck.

[Brian Drysdale]

Hi Andrew,

Thanks for the info, interesting that you currently recommend standard gamma.

I shot with the tests with the detail on MIN, since I normally do switch the detail off on HD.


Brian

[David Parks]

Originally Posted by Andrew Young

1) turn the detail setting down low, I recommend MIN or OFF. This is very important. The default detail setting is way too high for the big screen; 2) do not use the camera's Filmout gamma setting if you are shooting in a compressed format (tape or firewire). It is not useful in an 8-bit environment. 3) Do not use black compress. If you want a crushed black look, do it in post. 4) turn motion smoothing off and be sure white clip is set to 108%. Feel free to experiment with black stretch and knee. If you like the look of Cinelike, use it in the color matrix only, not in the gamma.

Based on everyone's combined experiences, I think we'll do a few shots with these settings in various scene file, the Stephen NOE's Panamatch and Tim Dashwoods "warm".

We'll try a couple of shots using the film out setting since we're bypassing HDV and going out analog component to Cineform.

Cheers

[Steve Mullen]

Originally Posted by Andrew Young

Anyway, I'll try to do a formal posting on this after NAB. What sort of data would you like to see to be convinced?

Since I used to live on 54th st. I've been to several of your filmout showings.

Film labs seem to fall into two classes: those that will take a tape or disk and simply burn it to film. This is the cheapest path. My assumption was FILMOUT was designed for use in these situations.

In talking with DuArt folks, I got the strong sense that while you could do you this -- the approach you favor is to go back to the original elements and process each, and then recombine them into a source to go film. In this case, I assumed -- perhaps wrongly -- that FILMOUT was less of a need.

But, you just told me you tried both -- and FILMOUT doesn't work for either case.

Which suggests that the curve designers didn't take into account either the JVC MPEG-2 encoder or the properties of any MPEG-2 encoder.

Thankfully, I guess, the need for a special monitor has kept many from trying this approach.

Curious, if many folks use the Panasonic setting with DVCPRO HD?

Also curious why you say "8-bit" since I think ALL recording formats use 8-bits. Why does it work with a Varicam, but not an HD100?

[Andrew Young]

Originally Posted by Brian Drysdale

Thanks for the info, interesting that you currently recommend standard gamma.
Brian

Hi Brian,
The only reason for this is that in my preliminary tests Cine gamma seemed to be pushing things together a bit. The colorist I showed it to felt she had a little more range with standard. However, I did not vary the level setting, as Tim and Paolo have done, which likely changes things. So please don't take it as a firm recommendation, just a comparison of those two out of the box settings. Tim and Paolo have really put much more analysis into this question of how to preserve the most information for post. What I hope to do is test how their settings hold up in the color correction suite and output to film. I'll keep you posted...

[Andrew Young]

Originally Posted by Steve Mullen

Film labs seem to fall into two classes: those that will take a tape or disk and simply burn it to film. This is the cheapest path. My assumption was FILMOUT was designed for use in these situations.

In talking with DuArt folks, I got the strong sense that while you could do you this -- the approach you favor is to go back to the original elements and process each, and then recombine them into a source to go film. In this case, I assumed -- perhaps wrongly -- that FILMOUT was less of a need.

Hi Steve,

Given the substantial cost of doing a filmout, it's hard to imagine the economic viability of shooting straight out to film with no color correction in the digital stage. This would leave all color correction to the film domain, not to mention the fact that you would have to telecine your negative (and possibly re-color correct) in order to have a color corrected video master. That hardly seems cheaper. In my mind, for a camera recipe to be viable it has to work well in the DI process - where you preserve as much information as possible for your DI color correct and then make your film and video masters from there. So even if Filmout did work directly out to film, I'm not sure it would make sense for most users. Are you supposed to make a print before you can send a tape to Sundance? I'm not sure that's practical.

Originally Posted by Steve Mullen

Curious, if many folks use the Panasonic setting with DVCPRO HD?

Also curious why you say "8-bit" since I think ALL recording formats use 8-bits. Why does it work with a Varicam, but not an HD100?

The two are very different animals, even though they may appear to have some similar qualities when viewed uncorrected.

I have done some shooting with Film Rec mode on the Varicam and I think it is a good system for improving dynamic range and maximizing options in post. However, it is not the same thing as JVC's FILMOUT setting. The latter, according to JVC, has no gamma correction applied at all, which I believe is a problem at shallow bit depths because there are not enough bits dedicated to sampling the darker half of the tonal scale, which is where most of the picture information ends up with this setting. The Varicam, on the other hand, does apply a gamma correction to the data, which distributes it more efficiently for sampling. At least, this is what I believe is going on. Back in the days when I shot only film, I didn't think about gamma at all. So it's a brave new world for me and I do not want to pretend to be an expert. All I can say with certainty is that I could not make the FILMOUT setting look usable either way.

The HDV image looks great when it is well exposed, but you do not have the ability to pull things out of underexposed areas as you do with some other formats. When I tried try to build a pleasing gamma into material shot with the FILMOUT setting I got tons of noise, because the information just wasn't there anymore.

[Brian Drysdale]

Originally Posted by Andrew Young

Hi Brian,
The only reason for this is that in my preliminary tests Cine gamma seemed to be pushing things together a bit. The colorist I showed it to felt she had a little more range with standard. However, I did not vary the level setting, as Tim and Paolo have done, which likely changes things. So please don't take it as a firm recommendation, just a comparison of those two out of the box settings. Tim and Paolo have really put much more analysis into this question of how to preserve the most information for post. What I hope to do is test how their settings hold up in the color correction suite and output to film. I'll keep you posted...

Andrew,

I'm hoping to be looking at our tests in an editing suite this week, so I'll be able to see the effects of their different setups in more detail.

Brian