Home of Professional Cinematography since 1996
First Looks At 24P HD To 35mm
class="Body">I've used 24P HDCAM for transfer to 35mm for an indie film...so anyway, I saw three shots from my 24P HD feature scanned out to 35mm. I didn't pick the three shots, so I didn't see a good cross-section of types of shots, plus they were all from Day One of the shoot.
class="Body">First shot was a wide shot of a smoky bar interior with colored lighting and visible heavily color-gelled fluorescent tubes in the frame. Used the Canon HD 5.5 to 50mm zoom at f/2.8 with a #1/8 ProMist filter.
class="Body">As with all the other tests of 24P HD to 35mm that I've seen, there is an inherent softness to the projected image -- that's ALWAYS my first impression even on stuff other people have shot. On top of that, the smoke & the #1/8 ProMist & the wide-angle shot added a little more softness. (I was using the ProMist NOT because I thought that HDTV or video in general needs to be diffused, but because I wanted some flaring from small light sources in the frame.)
class="Body">I also think that there was some slight back-focus problems with my zoom, not visible in the eyepiece, occasionally BARELY visible in the big HDTV monitor when shooting at wide-open apertures (after much zooming in and out, day after day, not sure I was just seeing things, etc.) but slightly more visible on the 35mm projected image. I don't know what the future solution to this problem is other than shooting lens tests at the start of production and spending the money to scan it out to 35mm and project it.
class="Body">Anyway, the problem was not bad but annoys me, being a sharpness freak somewhat. Didn't get to see the longer-lens end-of-the-zoom close-up for that scene (but did a later scene) to see the difference on the big screen with the wide-angle shot.
class="Body">The second and third shots were from a dream/flashback sequence, shot at the long end of the zoom. The post house expressed some concern because I went a little over-the-top with the lighting, using HMI PAR's as super hot backlights through smoke, plus adding a #1/2 ProMist on top of that. I wanted a lot of burned-out areas, so I was only concerned about any technical problems from having areas over 100 IRE, but I wasn't trying to hold any detail in those areas.
class="Body">(I didn't get to see any "normal" shots done without smoke, like some of my normal day interior scenes. I assume those will look cleaner & sharper.)
class="Body">Some observations: the stuff I shot with the burned-out white areas look fine scanned to 35mm -- no video artifacts, other than one: around a point of light, the glow from a ProMist filter produces slightly concentric rings instead of smoothly feathered glow. I noticed this effect in the Panavision tests that I saw before production. Something to do with the transfer to DLT and up-rezzing to 2K. But otherwise, knowing that if I shot 35mm negative I could have pulled some detail out of the white areas if I wanted (which I didn't) the final image looked very much like film (maybe the heavy ProMist & smoke of this dream sequence helped give it a "look" that diverted your eyes from any tell-tale video artifacts.)
class="Body">It was in the "real world" bar scene that there were more video-ish qualities. White lights that burned-out seemed to look fine, more like film, but COLORED lights that were very "hot", close to burning out, took on some odd video-ish qualities that is hard to describe, almost solarized. So the green and magenta fluorescent tubes in the frame looked a little "odd" -- super saturated (so the color held on film and didn't burn out) but with subtle but weird solarized qualities, like slight bands of dark & bright in the glow. I actually liked some of the look and might color-time the bar footage later to have a funkier color quality, the way Chris Doyle shoots night stuff like in "Fallen Angels", which involves a lot of push-developing. Sort of go for a "hot & noisy" saturated look, like some sort of chrome slide film push-developed.
class="Body">The images were pretty and the director was excited, but the overall softness (which I expected) again reminded me that I was not shooting in 35mm but something slightly lower in resolution. Better in sharpness than Super-16 blown-up, and not grainy, but something a little softer than straight 35mm. Maybe I could have digitally "sharpened" the shots or turned some of the edge enhancement back on, but I really would rather have the softness than any video-ish edge enhancement artifacts on the big screen, which are just plain ugly to me.
class="Body">I also saw some test footage of mine cropped & scanned out to 2.35 35mm anamorphic (that stuff from Day One was scanned out flat and projected 1.85, even though we are framing for cropping to 2.35) and in terms of sharpness, it looked pretty much identical to the flat 1.85 version, despite throwing away some picture information. The only difference was that the quality of the anamorphic projector lens became noticeable where I was screening the test.
class="Body">David Mullen Cinematographer / L.A.
class="Body" >I don't know what the future solution to this problem is other than shooting lens tests >at the start of production and spending the money to scan it out to 35mm and project >it.
class="Body">David, as long as you're using the video-style lenses, this will not guarantee success. I shoot a lot of documentary video, and it's very easy for the backfocus to slip; and of course, if you ever change lenses during the shoot, you have to reset the backfocus anyway, so you're starting from scratch again. I'd check it at least on a daily basis.
class="Body">What kind of chart were you using to set up the lens? I believe there are some new charts made especially for HD (ZGC? I think...) which could make this easier - they really 'snap' into focus, which could help with your HD monitor.
class="Body">----- George Hupka
class="Body" >but COLORED lights that were very "hot", close to burning out, took on some odd >video-ish qualities that is hard to describe, almost solarized. So the green and >magenta fluorescent tubes in the frame looked a little "odd" -- super saturated (so >the color held on film and didn't burn out) but with subtle but weird solarized qualities, >like slight bands of dark & bright in the glow.
class="Body">Was it more like a kind of soft posterization effect than solarized, perhaps? I recently saw a digital projection screening of "Fantasia 2000" (I had a morbid interest in seeing this kind of projection). Though animation is not the best example for testing the technology, it looked very well and "film-like", but, precisely in the real life sequences I observed in the highlights the kind of effect I'm referring.
class="Body">Great feedback regarding your experiences with 24psf. Which post production facility did you use for on-line and/or delivery of the files to the film xfer facility?
class="Body">Which film xfer facility did you do the test with? And which facility will the producer use to transfer the entire electronic film?
class="Body">Do you recall where the detail level was set in the F900 rental camera? What rental house did you use for the camera package? What lenses did you use EC style or videostyle. What brand Canon or Fujinon? What focal lengths? Any concerns about breathing etc? Did you have a 48i high resolution HD monitor available for viewing on set if so what size was it? Did you use a light meter only or did you use a waveform and/or HD color monitor to decide exposure decisions?
class="Body">Did you shoot with matrix on or off? The SMPTE 240 version or a custom set up? Did you do any color correction to the 24psf footage prior to xfer to 35mm film?
class="Body">Would you shoot 24psf again given the choice and budget to originate on 35 or 16mm film for future projects?
class="Body">The lenses I used were the Canon HD 5.5-50mm and 7.8-144mm. Both were pretty sharp; the 7.8-144mm had better distance spacing on the barrel for focus-pulling, but also breathed badly, while the 5.5-50mm hardly breathed at all. I also had some Canon primes with me.
class="Body">SIM Video sent that camera down from Toronto to an L.A. office. E-FILM did the scanning to 35mm for the tests and will do the whole feature. I believe they are using CFI for the film developing & printing.
class="Body">I believe the HDTV monitor was 32", plus I had a smaller 9" one.
class="Body">For various reasons, I did not have access to a waveform monitor. Mostly used my light meter to rough in the lighting, then a combination of the zebras in the camera (usually set to show 70 IRE) and looking at the big monitor to set exposure.
class="Body">Detail levels / edge enhancements were turned off. Sound was recorded separately.
class="Body">Filtration was mostly the #1/8 ProMist. A few shots or scenes used heavier diffusion. Almost everything was shot at 0 db. A few daytime scenes were shot at -3 db. And a few shots were done at +3 db.
class="Body">Image was composed for cropping later for 2.35 : 1 'scope output.
class="Body">Will I shoot another project on 24P HDTV? Sure, if it's right for the project -- but I'm withholding final judgements until I see the 35mm print of the whole film completed & projected. I'd like apply what I've learned from the mistakes I made and see if I can do better next time. But honestly, I'm also ready to go back to 35mm... I've been DYING to try out the new Fuji F-400 and still haven't been able to...
class="Body" David Mullen
class="Body" Cinematographer / L.A.
class="Body">David Mullen Wrote :
class="Body" >"The lenses I used were the Canon HD 5.5-50mm and 7.8-144mm. Both were pretty >sharp; the 7.8-144mm had better distance spacing on the barrel for focus-pulling, >but also breathed badly, while the 5.5-50mm hardly breathed at all. I also had some >Canon primes with me."
class="Body">What do you mean exactly by the saying that the lenses "breathed"? Can you explain under what circumstances this happens?
class="Body">Thanks for the info
class="Body" >What do you mean exactly by the saying that the lenses "breathed"?
class="Body">You know, when you pull focus on a zoom lens and the lens actually zooms a little during the rack, like the image is "breathing" in or out. Some zooms do this more than others.
class="Body" David Mullen
class="Body" Cinematographer / L.A.
class="Body">I'd like to thank George Spiro Dibie, IP 600 President for hosting a screening at Universal of DP David Smith's "Starry Night" and the very first look at 24p transfered to 35mm, "Nicolas."
class="Body">First off, Mr. Smith's PAL DigiBeta transfer of "Starry Night" was impressive. Light years ahead of DV cam, the DigiBeta transfer was good enough that you focused on the story. It looked like 16mm bumped to 35mm. Occassionally we'd see a scene that aliased or the camera's detail level was a bit too high.
class="Body">If you don't get a chance to see "Starry Night," look for Spike Lee's new DigiBeta movie slated for release soon.
class="Body">Now to "Nicolas!" Our audience of Union 600 members were very impressed with EFILM's 24p transfer to 35mm of "Nicolas." Billed as the first 24p feature, the trailer projected onto the big screen looked terrific. Not one person could find fault. No alaising, no artifacts, just beautiful images. I was especially impressed with the fire scenes and the camera's 11 stops of latitude. Depth of field problems have been solved with post effects blur tools. I can see why Lucas believes in 24p.
class="Body">Latitude of 11 stops... for video that would be 8+ stops into the shadow details. :-) The future might be multiple, collimated CCD's at various ND's getting you shadow detail and hold the sun's corona too.
class="Body">These post blur tools are also of interest. Is the foreground rotoscoped and then background softened ? Wonder if that compresses better too.
class="Body">But it sounds labor intensive. Please elaborate.
class="Body">I keep hearing about this "11 stops" hype but as of yet have not seen footage nor test result data that reflect the figure. I personally do not believe the camera is capable of an 11-stop range. Side by side comparisons with film, A/B splits and alternate segments on demonstration tapes from Sony continually show either (1) less high-light handling capability or (2)okay highlights but less shadow detail than the matching film scene. If you carefully read between the lines in the Sony brochures and data sheets, I think my position is supported. I think they slyly obscure the fact by citing tape format specifications when it suits their purpose. Their literature refers to the "HDTV camera system" and "captured images" i.e., tape playback, when Sony makes these "11 stop" claims.
class="Body">Larry Thorpe, both live and in print, carefully uses phrases like "the 24P format" and "the HDTV system" when citing the effective dynamic range, but carefully avoids these claims in the same breathe with actual process imaged specs, the actual pictures coming directly out of the camera itself.
class="Body">I think (personal opinion here) that the truth of the matter is the digital tape format RECORDING the camera output is indeed capable of recording and reproducing a signal with an effective 11-stop range. It's just that the camera head itself is not making such a picture in the first place.
class="Body">Okay, flame away.
class="Body">Jim Furrer VGG Systems, Inc.
class="Body">I always thought that the rule of thumb was that the stop range was one less that the bit depth, ie 8 bit gives 6 stops, 10 bit gives 9 stops and 12 bit gives 11 stops.
class="Body">Now these are theoretical maximums and generally work out to one less in practise.
class="Body">What is the bit depth of an HDCam recording?
class="Body">We can take it from there. that's not allowing for pre-processing to limit the range in any way.
class="Body">Sorry that should be 8 bit gives 7 stops in ideal conditions.
class="Body">Perhaps such figures are simply useless or misleading. I usually ignore them -- they are more useful for engineers than cinematographers maybe. The usable exposure range depends on so many factors, including the gamma of the print stock to be used if intended for projection.
class="Body">People who shoot film negative for telecine transfer to tape have the widest range of exposures to play with, but that doesn't mean that all that detail from high to low will be present if they stuck a print off of the negative (especially with Vision print stock.) And when shooting film, just because there's detail in the negative doesn't mean that you will get acceptable results when making a print. For example, if a scene is severely underexposed, you still might see detail in the negative you can use -- but when you print it back "up" the results are so milky, grainy, and unattractive that the shot really isn't usable.
class="Body">So much depends on the shot itself. The 24P HD camera I used seemed to have excellent shadow detail but poor ability to handle bright areas. But if your shot had no hot spots or bright areas in it (or the bright areas were so tiny that it made no difference how much over 100 IRE they went) and was mostly a shot of a room with objects of mid to low range of brightnesses, you might look at the HD image and say "wow, it really has great latitude -- you can see detail in the dark wood or that dark green couch or under that desk." It might seem similar to film negative in its latitude, if not seem to have a better ability to see into the shadows (depends on the film negative you are comparing it to.)
class="Body">In one 24P HD shot that I had scanned out to 35mm film, one thing that struck me was how much more into the shadows I was seeing compared to 35mm neg printed. I was filming a bar with dark walls with no light on them, yet I was seeing detail under the bar stools, and on the black ceiling, and in every corner -- in fact, there were no black shadows (the fact that the room was smoked helped).
class="Body">If I shot on 35mm negative, that information would also probably be there, yes, but an ordinary print stock would have too much contrast to allow me to see that detail -- so it's more that the digital transfer process was allowing me to pull that detail and put it in a range that the print stock could see it. In other words, if I had shot in film negative and went through a digital intermediate process, I could have gotten similar results if not better. But compared to simple film neg-to-print, the HDTV-to-neg-to-print seemed to have more shadow detail.
class="Body">And like I said, if you didn't want to hold detail in overexposed areas because they are small in frame, then overall you'd have the impression that the 24P HD camera was giving you a similar latitude to film negative, if not MORE ability to see into the shadows.
class="Body">BUT if you had large areas in the frame that were very bright, or let's say, were filming someone lit only by a practical lamp with a shade on it, then you'd notice that the latitude of the 24P HD camera was much lower than film negative.
class="Body">So in practical terms, the latitude of the 24P HD camera is really dependent on the shot itself. But you CAN say that film is more flexible in handling a wide dynamic range from extreme bright to dark, which makes it more practical for the largest range of lighting conditions.
class="Body">But if I were doing a documentary to be filmed on city streets & back alleys at night with no lighting, I'd probaly record more usable low-end detail with an HD camera with the gain boosted than by using Vision 800T film stock push-developed. I'd just be living with the fact that bright areas (like headlights or neon signs) would be burning out to white faster. So it depends on what you practically need the camera to do.
class="Body">So when Sony goes around saying that the 24P HD camera has "11 stops" of latitude, it's not so much a lie (well, maybe) as a misleading or simply useless statement. Real life has more than an 11-stop brightness range in it, so the issue really is what is the range that you need to capture detail within, and what happens to those objects that fall outside of the range?
class="Body">David Mullen Cinematographer / L.A.
class="Body">I do not know if the SONY HD camera has 11 stops or not, and I guess the only way to find out would be carefully conducted tests.
class="Body">That said, I personally also do not believe it has 11 stops. There is an interesting SMPTE Journal article about the SONY HDCAM camera (I cannot look up in which issue, as I am in beautiful sunny Vienna currently), which states that the chip has an actual 11 stops range, but that in downstream image processing in the camera this is reduced to a smaller range. In my opinion this explains where the SONY marketing people got this 11 stop number.
class="Body" Marc Shipman-Mueller,
class="Body" Technical Representative Arriflex Corporation;
class="Body">I would say that the 11 stops thing is pretty much propaganda. It would be pretty much my decidedly un-technical understanding of it that the information is there on the tape but to get it at one end of the spectrum (in grading) you will loose some at the other.
class="Body">Just my two cents
class="Body">Daniel Villeneuve, c.s.c.