HDTV format comparison article by the EBU

[Arachnotron]

The EBU published a short Technical Review (http://www.ebu.ch/en/technical/trev/trev_308-hdtv.pdf) on a HDTV format comparisons at the IBC 2006. Nothing really shocking, but maybe of interest to some...

The 720p/50 format showed better image quality than the 1080i/25 format for all sequences and for all bitrates;
With decreasing bitrate in the compressed domain, the difference between the 720p/50 and 1080i/25 format became more marked;
The 1080p/50 format was rated equal or better than 720p/50 for the higher bitrates – the extent depending on the test sequence. However, 720p/50 was rated better than 1080p/50 at the lower bitrates
.

[SeeMoreDigital]

Thanks for the link :)

[DarkZell666]

I am personally very shocked (amazed is more accurate :D) by the resolutions that HD is bringing us, and seeing at what framerate current AVC implementations work on our current systems (look at x264 with all maxed-out options ... well below 10fps on many systems @ 720*576), I wonder in what format's we'll be supposed to back up our HD-DVD's in the near future oO

CRT is very forgiving with artifacts, but all those LCD screens that are available aren't at all. Every single macroblock shows up like a spot on uncle mike's nose ;) I suppose encoding habits will likely be shifted in another direction (the tradeoff between encoding features, bitrate and resolution will be much different from now, imho).

[2Bdecided]

The EBU are very pro progressive.

However, in the UK, both Sky and the BBC are ignoring them and using 1080i on their HD broadcasts/tests.

Cheers,
David.

[scharfis_brain]

here in Germany the broadcasters decided for 1080i, too.

But the AVC encoders they are using aren't that good at all (Bad realtime implementations with faulty inloop deblocking I guess).
They produce severe blocking in difficult areas.
I order to make this worse they only use bitrates of 8mbps AVG (ranges 2 mbps to 16 mbps).

[jggimi]

In North America, the majority of ATSC broadcast is 1080i, but there is some 720p. From my cable system, it doesn't matter what the channel is, I must tell the set-top-box (STB) to output 1080i or 720p for all HD channels, regardless what ATSC signal it is receiving.

I started out with the STB set to output 720p, since that is what my TV is, but lately I switched it over to 1080i, and let the TV make the conversion instead. I noticed a few artifacts with 720p output which did not show up when using 1080i. Since I no longer recall who was doing the broadcasting, I cannot determine if it was a conversion problem in the STB or not, but I assume so.

[scharfis_brain]

I remember reading somewhere that broadcasters who use 720p for broadcasting do internally work with 1080i nontheless...

[2Bdecided]

here in Germany the broadcasters decided for 1080i, too.

But the AVC encoders they are using aren't that good at all (Bad realtime implementations with faulty inloop deblocking I guess).
They produce severe blocking in difficult areas.
I order to make this worse they only use bitrates of 8mbps AVG (ranges 2 mbps to 16 mbps).

Oh dear!

The BBC test is 20Mbps, but the Sky Broadcasts are about 12-14Mbps. The difference in quality is easily seen, which suggests the encoders aren't very good (or the content isn't).

Cheers,
David.

[SeeMoreDigital]

Oh dear!

The BBC test is 20Mbps, but the Sky Broadcasts are about 12-14Mbps. The difference in quality is easily seen, which suggests the encoders aren't very good (or the content isn't).

Cheers,
David.It's a crazy situation isn't it... Given the source, I reckon most people on the forum could generate near flawless looking 1440/1920x1080 encodes using good 'ol MPEG-4 Part 2 - and at lower than 12-14Mbps too!

[Doobie]

One of my biggest gripes with digital movies at the theaters is that the 24fps standard was kept. It should have been double, or tripled. That would have helped greatly to keep the movie theater industry ahead of home theaters.

It's great if HDTV can be scaled to 50fps. I didn't know this was possible (and, it's not currently). Wouldn't that be ironic if the home picture quality of the home theater exceeded that of the theater, assuming that consumers could get 50fps material.

[jggimi]

The ATSC standard supports up to 60fps, with 720p

http://forum.doom9.org/showthread.php?t=76006

[SeeMoreDigital]

One of my biggest gripes with digital movies at the theaters is that the 24fps standard was kept. It should have been double, or tripled. That would have helped greatly to keep the movie theater industry ahead of home theaters. Interesting.... What is it about the 24 frames per second speed you don't like - Or anybody else for that matter!


Cheers

[int10h]

Actually, modern theaters use a shutter rate of 48 or 72 Hz. The DLP projectors used in digital cinema operate at 50 or 60Hz.

Increasing the frame rate is quite useless, since neither the illusion of movement nor the flicker effect are directly linked to the frame rate.

[foxyshadis]

Some projection tvs (and projectors) can already display 24p content at 48 or 72hz, with a strobe, for theater-style viewing. I have no clue if it makes any difference, never seen it myself.

You're never going to convince production houses to move past 24 fps unless there's a major compelling reason - that's twice as much film to deal with, and it would probably change the 'film effect'. Maybe if major studios go 100% digital they'll get more flexible, but doubtful.

[2Bdecided]

Interesting.... What is it about the 24 frames per second speed you don't like - Or anybody else for that matter!


Movement s t u t t e r s

Fast pans are impossible


I hate 24 frames per second for the same reason I hated the 15 frames per second footage from my old digital still camera's "video" mode - it creates nasty movement artefacts that are not present in real life!

Cheers,
David.

[SeeMoreDigital]

Movement s t u t t e r s

Fast pans are impossible

I hate 24 frames per second for the same reason I hated the 15 frames per second footage from my old digital still camera's "video" mode - it creates nasty movement artefacts that are not present in real life!To confirm, are you referring to what you see at the "cinema" or at "home" on your TV?

[laserfan]

Movement s t u t t e r s

Fast pans are impossible


I hate 24 frames per second for the same reason I hated the 15 frames per second footage from my old digital still camera's "video" mode - it creates nasty movement artefacts that are not present in real life!David you're exactly right--I have said many times that it's too bad David Trumbell's "Showscan" system (60fps 65mm film) was not adopted. We are instead stuck with converting from ancient 24fps motion picture film to (any of the) video formats which are a compromise of it, with extra frames thrown-in and the resulting "judder effect".

I understand that HD-DVD movies are encoded as 1080/24p. So a purist (or anyone especially susceptible to judder) will want a 24p player and one of the new projectors that accept 24p input. Finally, source-to-screen without Telecine twitchyness!

[Doobie]

Interesting.... What is it about the 24 frames per second speed you don't like - Or anybody else for that matter!

As 2BDecided pointed out, fast motion stutters (strobes) at 24fps. If you're at the theater and there's a rapid pan, often the picture appears to jump rather than flow.

If a car races passed the screen, you'd see it appear at point A, then it would jump to point B, and then disappear. Movie makers work around this with motion blur. The blurring is less noticeable than the jumping.

I understand 24fps with old film. But, as cheap as digital storage is, digital filming should use a much higher FPS. The final product should be compressible to almost the same file size because there's much higher redundancy between frames.

[scharfis_brain]

24 or 25 fps have become an artistic effect, too.

an identical scene preseneted once in 24fps and then in somthing >= 48fps will have a much more impressive effect in 24 fps.

[2Bdecided]

To confirm, are you referring to what you see at the "cinema" or at "home" on your TV?

Both.

In the cinema, fast pans give rise to stuttering, strobing, double/tripple images or just a blur. AFAIK this depends on how the sequence was shot (especially shutter speed, depth of focus, and the content itself!), how many blades are on the projector, and whether your eye tracks the movement or just lets it go past.

On TV, you never get tripple images (or even double ones, unless you're watching on a huge interlaced display, in which case you can sometimes see combing if your eyes tracks the movement across the screen) but you can get all the other problems.

(I'm not talking about 3-2 pulldown here which, as I'm in a 50Hz country, is not something I'm used to. I have seen it on TV in the USA, and thought "how can people watch that - it's like it's starting and stopping the film several times a second!". Mind you, I dislike the pitch increase on the soundtrack on our 24>25 speed-up, but at least the picture doesn't stop-start).

(You can, of course, get strobing even on 50 or 60fps video, if you set the shutter speed far too high on very fast motion - but with care I think 60fps is sufficient. The only problem that remains is when there's fast motion which your eye can track while the camera does not - in this instance, I don't think any current frame rates are sufficient).

Cheers,
David.

[2Bdecided]

24 or 25 fps have become an artistic effect, too.

an identical scene preseneted once in 24fps and then in somthing >= 48fps will have a much more impressive effect in 24 fps.

Ah, the filmic effect.

But is this because most of the expensive content we have been exposed to in our lives is at 24fps, while most of the cheap/immediate content we have been exposed to is 50 or 60fps?

Whereas if you go back 25 years here in the UK, you find the opposite. At around that time, making TV programmes (or parts of them) on 16mm film was cheap and easy, while making them on video was comparatively expensive/difficult (especially outdoor/location filming, because of the size of video recorders back then!). It was a real step forward when professional video recorders became more "portable" and programmes started to be all video. Suddenly older programmes (which used film inserts for all location work, to avoid lugging big/heavy video cameras and recorders around) just looked old and crappy (picture quality-wise).

Mind you, 24p is easier to code, and the stuttery motion hides some coding artefacts which would be quite visible crawling over the smooth motions of 50 or 60fps.

Cheers,
David.

[int10h]

Whatever the frame rate, if you pay attention, you will notice motion artifacts, because the illusion of motion is due to psycho-visual effects in your brain. There will be stroboscopic effect, there is no way to get rid of it. The only thing we can avoid is the flickering, by raising the display rate (that does not have to be equal to the recording rate).

You can experience illusion of movement at a very low frame rate. It mostly depends on the characteristics of involved objects and of the background.

[MfA]

I know it's standard in film, but upping the display rate without upping the frame rate has it's downsides. It makes motion judder worse. Or at least that's what they usually say when promoting motion compensated frame rate conversion.

The world would have been a better place if 1080i had simply not made it into any of the standards ... 720p50/60 and 1080p24/50/60 are the only ones we really needed, but they had to fuck it up and cause "bigger numbers are better" we have to go through another generation of interlaced suffering.

[scharfis_brain]

upping framerate by adding an uneven number of frames(see 3:2 Pulldown with NTSC) makes things worse.
But upping the framerate with an constant number of frames (see 2:2 Pulldown for PAL) leaves motion smooth.

So it is a matter of the TV-Standard or the display.
Better displays are able to dynamically switch between 72 for film and 60 Hz for video if 3:2 pulldown is implemented. Or even 120 Hz. So motion remains fluent without any twelve Hz 3:2 Pulldown judder.

[MfA]

But upping the framerate with an constant number of frames (see 2:2 Pulldown for PAL) leaves motion smooth.
It leaves it familiarly unsmooth, similar to film, but smooth it ain't. It's stop and go from our vision's point of view if the refresh rate is low. If the refresh rate is high, and we track something on the screen, we track the average motion. The repeated frames then blur together even though they should not. You either get judder or blurring.

[scharfis_brain]

But with standrad NTSC Equipment you'll get addtitional judder, caused by the 3:2 pattern.

[KoD]

1. How much does it cost to process 1m of film ?
2. How much more film would have to be shot if you would capture 60 fps instead of 24fps ?
3. Are you willing to pay the price increase for a DVD when it comes out in stores ? (think 100$ per DVD at least)

a. Do you know why film is still used instead of recording on digital cameras ?
b. Can you spot the differences between a movie shot on stock 35mm film and one shot on a digital camera ?
c. Do you know why digital camera makers are trying to emulate the "35mm film" look ?
d. How much do you know about how the human brain percieves motion, distance, day and night ?

There's more in it than just resolution numbers and frames per second. And btw, 1080 is ridiculously low to what a film scanner can do.

[darkavatar1470]

Better displays are able to dynamically switch between 72 for film and 60 Hz for video if 3:2 pulldown is implemented. Or even 120 Hz. So motion remains fluent without any twelve Hz 3:2 Pulldown judder.Oh yeah, 120hz TVs to go with 120fps AVIs, life's good.

my eyes are very sensitive to flickering, so watching movies in a theater with a 24fps shutter makes my head ache. Too bad most of them still is.

[Doobie]

1. How much does it cost to process 1m of film ?

Probably nearly the same regardless of frame rate, and a fraction of what it cost just a few years ago.

3. Are you willing to pay the price increase for a DVD when it comes out in stores ? (think 100$ per DVD at least)

ROTLMAO

a. Do you know why film is still used instead of recording on digital cameras ?

Old habits. Filming with digital is cheaper and easier, for equally good results - and for better results if the digital film standard were more aggressive, but they only wanted to match the quality of film.

There's more in it than just resolution numbers and frames per second. And btw, 1080 is ridiculously low to what a film scanner can do.

What does that have to do with the price of tea in China?

HDTV is sufficient resolution. Increasing the resolution beyond that adds practically no visual improvement, but costs go up dramatically.

[*.mp4 guy]

@ doobie you can't match 35mm film quality with any curent digital cameras, you can't really even get close right now. Digital cameras suffer from very poor dynamic range, a tendancy to miss details that film cameras catch (this doesn't have anything to do with resolution, it has to do with the current limits of the ccds in cameras) all current "digital" cameras also suffer from either lossily compressed video, or an extra d/a-a/d conversion when their output is captured via raw component video, there are no digital HD cameras available that can keep the video in a lossless digital format, all of them either output an analog signal or compress lossily due to current bandwidth restrictions, which incidently makes digital storage cost just about as much as film storage if you don't use lossy compression.

Digital recording will catch up to analog at some point in the future, but its undeterminable if it will ever completely surpass it. Analog audio recording is still in wide use, both amoung consumers and profesional recording studios , mostly becuase of cost issues (for consumers) and lack of benefits for studios. Imo lossily compressed digital isn't really any better then analog anyway, either way you get a loss whenever you switch formats, either way the video is irreparably damaged. As long as film has better quality then digital cameras It will probably be the best solution.

[reepa]

Digital cameras suffer from very poor dynamic range, a tendancy to miss details that film cameras catch (this doesn't have anything to do with resolution, it has to do with the current limits of the ccds in cameras)

That's not true. Check this out:
http://www.clarkvision.com/imagedetail/film.vs.digital.summary1.html

The chart (under "Dynamic Range and Signal-to-Noise") shows how much worse two different films perform against a Canon digital camera.

Direct link to the chart:
http://www.clarkvision.com/imagedetail/dynamic-range-tfcn-film-1-600.gif

[*.mp4 guy]

You do realise that I was talking about motion video digital cameras right? Motion video cameras top out at 1920*1080*30P with about 8 bits of color depth (actually a bit less due to compression and a/d losses) Wheras Top of the line digital SLR cameras have ~4 times the spatial resolution and well over 4 times the color depth and also usualy support lower iso settings and have better signal to noise, they all also support lossless digital storage of pictures all of which gives them a HUGE edge in image quality, DSLR and Digital Film are really not comparable quality wise atm. All of that said that comparison was significantly less then completely thorough, there are films with better performance then those used for the chart comparison, and most digital motion video only has 5.5 (http://en.wikipedia.org/wiki/Stops_%28Dynamic_Range%29) stops as compared to films 11 (both being averages, there are better and worse performers in both categories of course)

(Edit) Forgot to mention that the comparison was inherently flawed since the film obviously had to be put through a film scanner before being analyzed, this process would inherently be limited by the quality of the scanner, which relies on the same technology as digital cameras and would unavoidably introduce generation loss even if it was able to significantly outperform the film, which is unlikely.

[Doobie]

I found this information about digital films:
RESOLUTION: 2K (2048 x 1080), 4K (4096 x 2160 pixels)
FRAMES PER SECOND: 24 fps or 48 fps (only possible for 2K)
COLOR DEPTH: 12bit/color = 36 bit total for RGB (68 billion colors)
IMAGE FORMAT: MXF (Material eXchange Format)
IMAGE BANDWIDTH: maximum 250 Mbps which is 112 GB/hour
AUDIO FORMAT: up to 16 channels of 20-bit 48 KHz or 96 KHz WAV
SUBTITLE FORMAT: PNG files or Timed Text
TOTAL MOVIE SIZE: between 45 and 140 GB/hour

In establishing these standards, they concluded that humans can only distinguish up to 11-bit color depth. So, using 12-bit plenty.

[foxyshadis]

The guy has a 6000 DPI drum scanner available, though that isn't necessary for movies. Something pointed out is that digicam sensors only have ~half the luminance resolution of their sensor size, and ~quarter color resolution, because of the RGBG sensors. Also pointed out is how bad film grain wrecks detail, though.

Edit: Actually, it seems to be:
GRGR
BGBG etc.
Which makes a little more sense. Same resolution issues apply.

The far more interesting comparison was another (somewhat more dated page) in which 4x6 film is compared to 35mm, proving how limited 35mm maximum resolution is compared to higher film sizes. 60 or 70mm (IMAX) will be able to outcompete digital for some time, probably 3-5 years at least, but if something like a 35mm equivalent is all that's desired, it seems digital can handle that fine. (For a much higher upfront cost than film cameras, presumably.)

[MfA]

(Edit) Forgot to mention that the comparison was inherently flawed since the film obviously had to be put through a film scanner before being analyzed, this process would inherently be limited by the quality of the scanner, which relies on the same technology as digital cameras and would unavoidably introduce generation loss even if it was able to significantly outperform the film, which is unlikely.
Using digital as an intermediate is already common regardless of the final medium of the movie.

Googling around a bit I see Return of the King was done in 2K for instance.

[*.mp4 guy]

I realise that, I was just pointing out a problem with the methadology, I understand that there isn't any solution to the problem, but that doesn't make it any less relevant. The performance of film scanners is quite good, but in a comparison such as this there limits can cause prblems.

[reepa]

(Edit) Forgot to mention that the comparison was inherently flawed since the film obviously had to be put through a film scanner before being analyzed, this process would inherently be limited by the quality of the scanner, which relies on the same technology as digital cameras and would unavoidably introduce generation loss even if it was able to significantly outperform the film, which is unlikely.

The scanner is hardly a limiting factor. It can resolve way more detail than available on a 35mm film:
http://www.clarkvision.com/imagedetail/scandetail.html
Neither is the noise from the scanner a limiting factor:
http://www.clarkvision.com/imagedetail/dynamicrange2/
Especially this image:
http://www.clarkvision.com/imagedetail/dynamicrange2/compare-shadows.v1.jpg
"The shadow detail in the film was confirmed visually by inspecting the film under magnification: the scans show what can be detected by visual inspection plus a little more."

Something pointed out is that digicam sensors only have ~half the luminance resolution of their sensor size, and ~quarter color resolution, because of the RGBG sensors.

This of course doesn't apply to cameras with three sensors, or sensors without bayer filters (like Foveon X3). For example, according to Wikipedia, Panavision Genesis digital movie camera doesn't have a bayer filter even though it only has one sensor.

However, I don't know how much CCD and CMOS sensors differ between still photography and video applications. But the fact is that "digital" doesn't automatically mean worse than film.

Googling around a bit I see Return of the King was done in 2K for instance.

I wonder why such a big-budget production was done in "only" 2K resolution a decade after Snow White and the Seven Dwarves was restored in 4K resolution.

[*.mp4 guy]

At this time digital isn't yet a better solution then film, and a few anecdotes showing the exception to the trend aren't going to prove anything. The results you have sighted on that website are not inline with a lot of other results I've seen, 12 bit scanners can't capture the full dynamic range of film (at the low and high ends, one end usually gets truncated in a very high dynamic range image). That picture is close to useless since it is so low resolution it hides most of the problems that a scanner would introduce; I'm not saying scanners make a huge difference, but in the case of the original comparison a scanner would effect the results. Keep in mind, I'm not saying that scanners can't catch all of the detail in a scanned image (they can), I'm just saying that they can't avoid adding some defects to the image, allong with the detail.

In the detail results you sighted it is apparent that digital cameras can only more-or-less match 35mm film detail with ~16 megapixels (and can't match films color depth even in this case, compare this to digital films 2 megapixels and less then 8 bit color depth), it is also clear that it will be a long time until large format film is challenged by digital cameras.

[MfA]

Whether digital intermediates can or can't capture the full dynamic range is irrelevant, they are becoming standard either way. Dynamic range might have been an "advantage" of film in the past ... but with digital intermediates it isn't anymore. Only the practical downsides of going completely digital are relevant, not the theoretical.

[smok3]

i would think that DI is pretty much a suggested work flow by now, and the work flow can actually hide all kind of bad stuff the motion film may produce by itself (like with color corrections for example), MfA is right.

anyway to skip the bayer pattern, i would really like to see some info about 24fps and its psyhovisual values? links please :)

p.s. *.mp4 guy, leica 35 still is bigger than 35 motion film still (aprx; 36x24 vs 24x16 mm).

[reepa]

At this time digital isn't yet a better solution then film, and a few anecdotes showing the exception to the trend aren't going to prove anything.

"A few anecdotes" qualifies as disproof by counter-example to the claim that "you can't match 35mm film quality with any curent digital cameras, you can't really even get close right now", which initially prompted my response. I wasn't thinking in terms of digital video only. You're probably right about digital video.

The results you have sighted on that website are not inline with a lot of other results I've seen, 12 bit scanners can't capture the full dynamic range of film (at the low and high ends, one end usually gets truncated in a very high dynamic range image).

Could you link me to some of them? I'm very interested.

That picture is close to useless since it is so low resolution it hides most of the problems that a scanner would introduce; I'm not saying scanners make a huge difference, but in the case of the original comparison a scanner would effect the results. Keep in mind, I'm not saying that scanners can't catch all of the detail in a scanned image (they can), I'm just saying that they can't avoid adding some defects to the image, allong with the detail.

What kind of artifacts does the scanner introduce, then? The guy who did those comparisons confirmed the results from his scanner visually with a magnifier! I still don't think the scanner would introduce large enough artifacts to tip the balance in favor of film in those graphs I initially linked to.

(and can't match films color depth even in this case

I still dispute this, in light of R.N. Clark's results I linked to.

Also, I need to reply to an earlier post of yours:

all current "digital" cameras also suffer from either lossily compressed video, or an extra d/a-a/d conversion when their output is captured via raw component video, there are no digital HD cameras available that can keep the video in a lossless digital format, all of them either output an analog signal or compress lossily due to current bandwidth restrictions, which incidently makes digital storage cost just about as much as film storage if you don't use lossy compression.

There are digital cameras that can output uncompressed high-definition video. The video is recorded on a hard disk array for example. According to Rick McCallum, the digital tape they used to record Star Wars Episode II on cost over a hundred times less than an equivalent amount of film.

[*.mp4 guy]

Do a little research into profesional HD cameras, they recorded starwars by digitizing the cameras component video output (I know this because the camera they used does not have lossless digital output, only lossy digital or component analog video).

A while ago I looked into buying a film scanner (and decided it wasn't worth investing in when digital will eventually be a better solution). The Nikon Super Coolscan 8000ED Retails for more then $2000 and scans at 4000dpi with 14 bits per channel, it takes 4 minutes to scan a singal piece of film (more or less depending upon film format). It also supports a function where it scans the film multiple times and then averages the results to gether to reduce sensor noise, in this mode it takes about 16 minutes to scan one piece of film. This is a scan withought multisampling (http://www.imaging-resource.com/SCAN/CS8K/C8TRAIN7.HTM), and this a scan with multisampling (http://www.imaging-resource.com/SCAN/CS8K/C8TRAIN9.HTM). Draw your own conclusion.

[reepa]

Do a little research into profesional HD cameras, they recorded starwars by digitizing the cameras component video output (I know this because the camera they used does not have lossless digital output, only lossy digital or component analog video).

I used the Star Wars movie as an example of the cost of tape vs film, not as an example of uncompressed digital recording. Panavision Genesis for example outputs uncompressed digital (HD-SDI). Also, Dalsa Origin (advertised exposure latitude of 12 stops!) outputs 4k digital (over 4 InfiniBand buses). Arri D-20 outputs 2880 by 2160 at 12-bits.

A while ago I looked into buying a film scanner (and decided it wasn't worth investing in when digital will eventually be a better solution). The Nikon Super Coolscan 8000ED Retails for more then $2000 and scans at 4000dpi with 14 bits per channel, it takes 4 minutes to scan a singal piece of film (more or less depending upon film format). It also supports a function where it scans the film multiple times and then averages the results to gether to reduce sensor noise, in this mode it takes about 16 minutes to scan one piece of film. This is a scan withought multisampling (http://www.imaging-resource.com/SCAN/CS8K/C8TRAIN7.HTM), and this a scan with multisampling (http://www.imaging-resource.com/SCAN/CS8K/C8TRAIN9.HTM). Draw your own conclusion.

The Nikon is a CCD scanner. I assumed the data for the graphs on clarkvision.com was done with a drum scanner, since foxyshadis said the guy has a 6000dpi drum scanner and there are drum scans on his website. However, I can't find anything on that particular page confirming that, so I guess I'll have to e-mail him and ask.

To compare a drum scan to a ccd scan, check out this page:
http://www.clarkvision.com/imagedetail/scandetail.html

Direct links to images:
http://www.clarkvision.com/imagedetail/resol_exampl1.jpg
http://www.clarkvision.com/imagedetail/scanarea1-spnkhp.jpg

I do believe a CCD scan would corrupt film's performance in the signal-to-noise / dynamic range comparison.



An on-topic comment:
owever, 720p/50 was rated better than 1080p/50 at the lower bitrates
I hope better optimized compressors will improve the situation for 1080p in the future. Fortunately, Bluray and HD-DVD have ample bandwidth for 1080p.

[*.mp4 guy]

=/ When I researched film scanners I didn't find anything about drum scanners, I suppose this was probably because I was more interested in motion picture film scanning, where drum scanners are irrelevant (at least I didn't find any motion picture drum scanners). After doing some research, if he did use a drum scanner for the comparison, then the scanner is most likely a negligible influence on final quality.

I see that I am a bit behind on high end digital cameras aswell, the Dalsa looks like it could become a good film substitute (though storage costs would increase quite a bit, it would probably still be cheeper then film, if only because you don't have to keep the out-takes). Hopefully the next all digital films won't have the horrible visual quality of the new star wars films, its ironic but eps 5 and 6 are probably better looking then 1 2 and 3.

[foxyshadis]

Except that they're so lousily transferred, over and over, but otherwise yes. But Lucas lavished money on them; whenever money is tight, especially for tv stations, documentaries, concerts, and so on, budget digital is probably better than budget tape. Especially since it'll all be transferred for editing, as mentioned.

I don't know enough about the top-end digital movie cameras though. I'm not sure who the big manufacturers are. It's an interesting topic.

[smok3]

The new products that will be very interesting in the near future are coming out thought, they are marketed as 'lots of bangs for medium amount of money' (like red camera or silicon imaging/cineform combos). Curently especially products that are native cineform look very interesting for indie type of production, check;
http://cineform.blogspot.com/2006/07/silicon-imaging-in-48-hours-success.html

[*.mp4 guy]

Damn wmv, I can't tell how good the video quality actualy is with it spewing bands all over the place; there is no way a 10 bit raw would have that much banding (unless the camera is crap and the extra bits are only used for HD ccd noise retention), it looks more like 5 bits then 10. the video performace was pretty good otherwise, good detail retention when the lense wasn't killing the picture - there really needs to be wider support for custom lenses on digital movie cameras that don't cost as much as a house.

[movmasty]

The guy has a 6000 DPI drum scanner available, though that isn't necessary for movies. Something pointed out is that digicam sensors only have ~half the luminance resolution of their sensor size, and ~quarter color resolution, because of the RGBG sensors. Also pointed out is how bad film grain wrecks detail, though.

Edit: Actually, it seems to be:
GRGR
BGBG etc.
Which makes a little more sense. Same resolution issues apply.

The far more interesting comparison was another (somewhat more dated page) in which 4x6 film is compared to 35mm, proving how limited 35mm maximum resolution is compared to higher film sizes. 60 or 70mm (IMAX) will be able to outcompete digital for some time, probably 3-5 years at least, but if something like a 35mm equivalent is all that's desired, it seems digital can handle that fine. (For a much higher upfront cost than film cameras, presumably.)

IMAX shoul have a vert res of 24.000 pixels, so i guess 35mm is 12.000.....compete what?

1080 is a tv res, not a movie res.

[Atamido]

From the Wikipedia article on IMAX (http://en.wikipedia.org/wiki/IMAX): The intent of IMAX is to dramatically increase the resolution of the image by using much larger film stock at a resolution of about 10000 x 7000 pixels. To do this, 70 mm film stock is run "sideways" through the cameras. While traditional 35 mm film has an image area that is 48.5 mm wide and 22.1 mm tall (for Todd-AO), in IMAX the image is 69.6 mm wide and 48.5 mm tall.
That would make 35mm film about 6968 x 3190, although I would suspect that the reality of most 35mm film is much lower than that in a practical sense. The 4K (4096 x 2160 pixels) quoted above seems like a much more reasonable number for what might be possible. Though I wonder what resolution nice digital theaters project at. 1920x1080 seems much to low for a real theater.


I know it's all about personal tastes, but I personally prefer to view clean/smooth video. I know people talk about "grain" and such, but usually it just annoys me when watching movies. Same goes for the low framerate. I would prefer if the framerates were increased to 48 or even 60 fps for truly smooth video in fast pans and such.


It is worth mentioning the different methods out there for creating frames using different types of motion estimation. Depending on your content, this technology can work really well, and results in a very smooth looking video. I expect that at some point in the future some projectors/displays will be equipped with this internally to be able to create extra frames to make the video more smooth.

[*.mp4 guy]

I find that 24 fps pans only look choppy when I view them on a non mod-24 herts display or when the shutter speed was set to high (you wouldn't think this would be an issue with big budget films, but it seems to happen fairly often). What does bother me while watching movies is the stuttering caused by the "refresh rate", of the projectors, they really should up it to 96 or 120 and get rid of the strobing.