F900 comparison with Super-16

[David Mullen ASC]

I got some frame grabs back from the HD / S16 test I did in Austin, TX. If anyone wants to volunteer some of their web space, I can post some of them here.

 

The S16 was 7218 (rated at 400 ASA) and it probably could be retransferred a little better (it's a bit darker than it should be), so don't take this as some sort of definitive test or that I tried to make the S16 look worse than the HD. The 7218 was shot on an Arri-SR3 with an 11-110 Zeiss zoom; the HD was an F900/3 with a Fujinon zoom. The S16 footage was transferred to the same format as the HD camera, 23.98PsF HDCAM, on a Spirit.

[David Mullen ASC]

Looking at the frames on Paint Shop Pro, I think there's something wrong with the way the S16 was telecine transferred because it's WAY too grainy/noisy. Maybe if I post one of them, you'd see and tell me what went wrong... I may have to have the footage shipped to LA and redo the transfer if I want to be fair, but of course, I'd have to find a place willing to do it for free.

[J. Lamar King IMPOSTOR]

I emailed you Dave.

[David Mullen ASC]

Here's the two gray scales shot at the head of the 7218-to-HDCAM and the F900/3. As you can see, the transfer of the 7218 is unusally noisy so hopefully I can get it redone somewhere. Does speak though to the problem of being at the mercy of the labs and telecine people to show film at its best.

 

I reduced these photos by 50% but we'll see what size they show-up here as.

 

The F900/3 was shot with Detail "off".

 

 

 

Thanks J...

[David Mullen ASC]

Here's an interesting low-light test. Both cameras rated at 1000 ASA, shot at T/2.2 on both cameras. The 7218 was pushed one stop and the F900 was boosted to +3 db and the shutter changed from 1/48 to 1/32 (+3 db is a half stop gain and so is the change to 1/32, so a one-stop gain total.)

 

Thanks again to J. Lamar King.

 

 

[Alvin Pingol]

Thanks for posting those!

 

On the second set of shots, I was surprised at how well both formats held up to the one stop push. Mr. Mullen, if you were to shoot HD and needed a one stop increase in sensitivity, would you use the below method (half and half) or +6dB gain instead? From the image below, +3dB doesn't seem to add objectional noise (though these images are at 50%, so I suppose that would considerably smooth things out), and +6dB would allow you to retain the 180 degree shutter motion rendering. To my eyes, there is an obvious difference between 1/48th and 1/24th shutter speeds - not sure if 1/32th would be so easy to spot. What did you think?

 

As for image characteristics, it almost seems as if the 16mm burns out to white "sooner" than the HD shot (see candle flame, out of focus Christmas lights), but the portion of the candle right below the flame appears to have a smoother dark-to-light transition on the film frame versus the HD frame, which almost looks as if there is a "chunk" of certain brightness before it progresses to the lighter/darker colors. I know my video camera does this too, especially when auto-knee is enabled. Was it on during these tests?

[Michael Nash]

I couldn't resist trying to get the HD contrast to match the 16mm -- especially on the face. This was with Photoshop on a still, so it doesn't really prove much. But it's the type of thing you'd do in color correction with HD anyway.

 

As you can see, there's a lot of information in the HD image to work with, but the abrupt transition into highlights causes some problems. You can see it in the candle.

 

Just for yucks, I added one pass of sharpening to the HD image as well. I also reduced both images about 40% to save space here, but the example is really meant to show the contrast in the images.

[Phil Rhodes]

Hi,

 

> but the portion of the candle right below the flame appears to have a smoother dark-to-light transition

> on the film frame versus the HD frame

 

Yes, very noticeably, presumably because the F900 is pushing its red channel into DCC - this is what everyone gripes about with video and as we see here the effect is very apparent.

 

What surprises me here is that I prefer the film for the low-light image and the video for the brighter one.

 

Phil

[David Cox]

Execellent post.

 

Would it be possible to get hold of an uncompressed still of each so we can see what the JPEG compression is doing to the image?

 

Like Phil, I am surprise at the highlight detail from the HD image. Traditionally, this was the weak point of digital cameras - although the 16mm image has been graded to be brighter in the mid-tones so maybe this is pushing the highlights up a bit? As David says, once the 16mm frame has an improved TK transfer, this will be an even better comparison.

 

Also the digital footage seems to be a little warm - another usual example of digital footage. Whenever we are asked to make "video look like film" (!!), I drop the blacks, raise the whites and take red out of the mid-tones before even looking at the footage!!

[David Mullen ASC]

For the low-light shot, I tend to prefer the HD image -- the shadow detail is a little more open (although the S16 probably was transferred too dark) and of course, it looks cleaner.

 

Yes, you're seeing the effects of three things: DCC (Auto Knee), Knee Saturation, and ITU-709 Color Matrix Preset. If I were shooting this scene for a movie, I'd probably turn some of those off, maybe the ITU-709 first.

 

Also, I used the 3200K preset white balance. This accounts for some of the warmth in the gray scale, although personally I think the film gray scale is timed too blue.

 

The shots were emailed to me as large jpegs -- they would have been too large as uncompressed RAW or BitMap files.

 

The tests do sort of show that HD is closer to S16 resolution but 35mm in terms of lack of grain.

[Tenolian Bell]

In the 16mm image that seems like video noise to me. The grain feels too even to me. The candle, his face the christmas lights, the background are all at different footcandles. And so the grain shouldn't be the same at each level.

 

From my experience in extreme underexposure film would block up mostly around high contrast areas around the highlights.

 

On the other hand if that really is film grain that means the image is underxopsed even beyond pushing it.

 

I've pushed '74 many times and there is no real noticeable difference. I've pushed '18 a couple of times, the sharpness begins to drop off, but other wise under proper exposure there is no significant increase in grain.

 

A couple of questions I have.

 

What lens were both shot with,

what telecine was used for the 16mm,

and to what tape format?

 

You shot the man and candle at 2.2, was that from his face or an average for the entire scene?

 

What conditions were the girl shot under?

 

To me looking in the shadows the 16mm is more open, but the candle is more defined in the HD. I think the 16mm transfer isn't as unbiased as it should be for a clear estimation.

[Tenolian Bell]

I have been really impressed though, how video can handle low light situations. Properly exposed of course.

[Preston Herrick]

Perhaps the colorist used little or no noise reduction in the transfer. The denser the negative, the more light they have pump through it. As the gain goes up, so does the noise - which requires more NR. With NR, you walk a fine line before the image turns to mush. Of course, NR in telecine can help manage 16mm grain.

[Felipe Perez-Burchard]

This is a great test, thanks for the generosity David & Mr. King.

 

Without ever having transfered on a spirit, 7218 (despite how much better it is from previous stocks), is still too grainy to me (for general use, if the story warrants it of course then that's cool) and looks like these stills.

 

I was curious what the "speed" of the particular F900 camera was, usually I've seen it rated at 320.

Maybe a fairer grain comparison (for the non low-light stuff I mean) would be betwen 200 ASA pushed one stop (which I think is not as grainy as 500 ASA).

 

Or is there a reasoning why this would be wrong? (I'm just a sponge here absorbing knowledge).

 

Thanks.

 

-felipe.

[David Mullen ASC]

I shot the candle shot at f/2.2 on both cameras' zooms (11-110 Zeiss on an Arri-SR3 and a 18x 7.8 Fujinon on the F900), basically wide-open. My meter was set to 1000 ASA and the face next to the candle read one-stop under if I exposed at f/2.2. I shot the F900 at the same f/stop, boosting the gain +3 db and increasing the shutter speed to 1/32nd. The 7218 was pushed one-stop.

 

The F900/3 ended up being rated at 500 ASA at 0 db with a 1/48th shutter, based on using a waveform monitor and an 11-step chip chart, setting the white chip at 100 IRE and the Zone 6 chip at 70 IRE. So at +3 db and 1/32nd, I was at an effective 1000 ASA. The picture just looked too bright at 320 ASA. I rated the 7218 at 400 ASA and basically just shot the F900 one-third of a stop closed down from whatever the Arri-SR3 was set at, except for the separately push-processed shot of the candle, where I just used the same f-stop on both cameras.

[Tenolian Bell]

Without ever having transfered on a spirit, 7218

<{POST_SNAPBACK}>

 

 

Here is a link to a music video (Mosh It Up) I shot entirely on 7218 transfered on a spirit to HDCAM.

 

 

 

http://teckmoney.com/videos/index.html

Edited December 23, 2004 by tenobell

[Landon D. Parks]

I have to say im would take the HDCAM images over the 16mm stuff.... I can handle a little highlight ptoblem, I cant handle 10 tons of grain though.

 

FWIW, I thought both the 16mm and HDCAM images looked exactly alike, except for the grain in the 16mm frame.

 

In fact, I prefer the look the HDCAM produced over the sort of "Washed Out and grainy" look of the 16mm transfer.

[David Mullen ASC]

I have to say im would take the HDCAM images over the 16mm stuff.... I can handle a little highlight ptoblem, I cant handle 10 tons of grain though.

 

FWIW, I thought both the 16mm and HDCAM images looked exactly alike, except for the grain in the 16mm frame.

 

In fact, I prefer the look the HDCAM produced over the sort of "Washed Out and grainy" look of the 16mm transfer.

<{POST_SNAPBACK}>

 

I think you're seeing both grain and a lot of noise -- I'm hoping I can get the transfer redone. It won't be grainless though but much better hopefully. Perhaps does demonstrate though how film is at the mercy of the labs and telecine people unless to monitor them closely. But I think I can match the two examples more closely. However, yes, Super-16 grain will always be bigger than normal HD noise levels. But you can see lots of Super-16 on TV these days that look fairly fine-grained, like the show "The O.C."

[Guest Andy Sparaco]

I think you're seeing both grain and a lot of noise -- I'm hoping I can get the transfer redone.  It won't be grainless though but much better hopefully. Perhaps does demonstrate though how film is at the mercy of the labs and telecine people unless to monitor them closely.  But I think I can match the two examples more closely.  However, yes, Super-16 grain will always be bigger than normal HD noise levels. But you can see lots of Super-16 on TV these days that look fairly fine-grained, like the show "The O.C."

<{POST_SNAPBACK}>

 

Your setups where really interesting but as I'm sure you know tha context is everything. A couple of stills from a static set-up is a starting point and folks without a lot of experience in f Film to tape xfer and Video to Video grading are gonna draw conclusions which are inaccurate. I'm sure you know David (Happy Holidays) :D

 

There is a large difference in perception of grain between a still image and a stream of motion pictures. The human eye is a powerful noise reduction unit

 

We have been shooting 7218 since it was introduced and with Spirit and even Cintel xfers have found only very mild noise reduction is necessary to reduce how obvious grain is. Once xfer to D-1, digibeta or D-5 HD and even the Varicam format is complete all of the digital tools come into play. Plat the footage into a inferno/flame/smoke and you can make it do anything

 

We have even used it for Greenscreen work on difficult locations

 

With the introduction of 7217 our agency clients are'nt saying why not shoot HD, they are saying why bother with 35mm.

 

Maybe a new thread should be posed "How soon will Super 16 replace 35mm?"

 

As always a properly exposed, in focus negative without damage is the starting point for a beautiful transfer.

 

The OC/Malcom in the Middle/Scrubs/The Shield(?) others...? as you said Lot's of Super16 on network. I'd say smart producers looking for a pay-off in future syndication. We work in a business where people do what works and film orgination has always worked.

[David Mullen ASC]

For television, maybe, but graininess issues are still relevant when talking about theatrical projection. Considering I wish we were still making movies in 65mm, the last thing I want to consider is replacing 35mm with Super-16 for theatrical work...

[Landon D. Parks]

I wish we where making 65mm films today too. And really we would be, if it was not for the extreme cost of doing so.

 

I have seen only one film shot in 65mm, but when comparing it with 35mm, I see a huge difference.

 

but graininess issues are still relevant when talking about theatrical projection.

Grain was VERY noticable (A lot) in Never Die alone. So much so it was annoying at some times!

 

If your shooting for theatrical, Super 16mm should only be used if you have too. 65mm is the best, 35mm is the norm, and HD is a 2nd choice down from 35mm, 16mm is down from HD (Due to its grain) in my opinion.

 

I am sure a lot fo grain can be reduced if you use a D.I instead of optical to tranfer the 16mm - 35mm.

Edited December 25, 2004 by Landon D. Parks

[David Mullen ASC]

Well, you can get a fine-grained blow-up from Super-16 for theatrical projection -- but it may mean working with the slow-speed stocks, that's all. This is a rather simplistic way of looking at it -- enough to be inaccurate -- but one way to think of HD is that it's like 50 ASA Super-16 but with a 500 ASA speed... and very poor overexposure latitude!

[Tenolian Bell]

Grain was VERY noticable (A lot) in Never Die alone. So much so it was annoying at some times!

<{POST_SNAPBACK}>

 

That was video noise and artifacting from the DI.

[Michael Nash]

It's not always just about the grain. I'm one of those people who doesn't mind a little grain at times (when it's approporiate), but I DO mind a soft image. Super 16 and HD seem almost neck-and-neck in this area, at least on a larger screen. I always come out of the theater rubbing my eyes after viewing a 2K DI film... :blink:

 

HD is no doubt a clean image (unless gain-boosted), even when projected on a large screen. But 35mm is and always has been in a class by itself.

 

I'm impressed with how nice the current 16mm and HD shows look on broadcast. For example The OC and Reba look fantastic. Then I switch channels to Law & Order and BOOM! The 35mm and sparkling transfer cut through like a knife...

[Landon D. Parks]

To tell you the truth, from what I have seen, I prefer the look of HD is low light conditions with no gain boost, but once HD crosses a certain line into material that is bright, HD turns from preferable to near horrible.

 

If HD and 35mm where at the same resolution, I would prefer to shoot HD for darker scenes and then shoot the 35mm for brighter scenes. Because personally, I think 35mm film cuts up once it gets to dark....