Best Quality workflow 1080p 30fps on blu-ray.

[Greg Voevodsky]

After a lot of reading about codecs, my head is spinning.

What I want to do is deliver the best quality 1080p at 30fps (which right now I do not see as possible in either FCP or Avid without the super high end codecs running on expensive machines.)

Right now, I see the best path at 1080p with FCP using REDCODE 1080p at 24fps is possible and hopefully 30fps and then using mpeg4 for blu-ray.

On Avid, I see right now, the best seems to be avid's codec DNxHD 220X MXF - 10 BIT 6:1 (220mb) at only 1080p 24fps (I'd like to shoot and edit 30fps) and then encode mpeg4 for blu-ray.

The other broadcast codecs are all less colorspace, and less pixels.

Is it the best to then choose either REDCODE 1080P AT 24 AND HOPE YOU will have 30fps by Nab or AVid's solution above.

Or what may be coming?

Blu-ray is definately going to be better than broadcast and masters should be saved on harddrive or holodrive in either redcode or AViDS code. Masters should then downrezed to broadcast in HDCAM or HDVCPRO. The 4k RAW footage should be kept on HD for future purposes.

Any other ideas or workflows?

ps - I hope REDCINE might be the best to take our master edit out to mpeg4 for blu-ray. I hate all the mpeg 2 artifacts I get with water right now, that was not on my DVcam original footage, even at the max encode rates!!!

[Graeme Nattress]

REDCODE doesn't care what frame rate you pick. It just encodes and decodes frames. If you're NLE supports a 37.5fps frame rate, then the codec would too.

Codecs that do 1080p30 are everywhere. You could, in FCP as that's what I know best, use:

uncompressed 8bit 4:2:2
uncompressed 10bit 4:2:2
Animation 4:4:4
PhotoJPEG 75% 4:2:2
PhotoJPEG 100% 4:4:4

etc. None of which need any extra hardware to work.

Any long GOP codec is going to have motion issues if the bit rate is not high enough. Nothing we can do about that, other than offer you the nicest, cleanest signal to work with.

Graeme

[Jeff Kilgroe]

After a lot of reading about codecs, my head is spinning.

What I want to do is deliver the best quality 1080p at 30fps (which right now I do not see as possible in either FCP or Avid without the super high end codecs running on expensive machines.)

I'm curious as to why you would want 1080p30? Most HD-DVD and BluRay features are authored at 1080p24 to match with the theatrical film source. Broadcasters, as they begin to migrate to 1080p, will be looking to do 60fps. Current 1080i at 60 fields (30 frames) per second is actually superior to 30p if you're filming fast motion, sports, news, etc... Really, the only reason I'd deal with 1080p30 is if I needed to deliver 1080i and I only had a progressive acquisition source without bandwidth or storage capacity to handle full 60p.

Other than that... Uh, just what Graeme said.

[Graeme Nattress]

Although I wouldn't say interlaced video is superior to anything. Think 24p, 30p and 60p. (and 25p and 50p) But BluRay and friends can't do 50p or 60p at 1080 (do they allow 720p60 - I hope so), so they're not exactly future proofed. I just don't see why with HD they couldn't have made a nice clean break away from our analogue (ditch the 1000/1001 NTSC factor in frame rates) and standard def (interlacing) past.

Graeme

[Stuart English]

But they didn't break from the 1000/1001 past, nor interlace either. Too much history and engineering to dig through there to worry about exactly why. Bottom line is HDTV broadcasting in the USA has two options - 720p/59.94 frames/ sec or 1080i/59.94 fields / sec. At a maximum of 22Mb/s in MPEG-2. Thats it. Anything else means a total rework of the DTV infrastucture.

So in this sense, 1080i/60 can certainly be considered better than 1080p/60 as the latter cannot be transmitted. And 1080i certainly also looks different to 720p/60 - some prefer it's higher spacial resolution and some don't.

What I'd be looking to create as an edited master would be 1080p/60, so that I can make a dub to 1080i/60 or 720p/60 for HDTV broadcasting and wait for when Blue-Ray, HD-DVD or internet or satellite based TV services - that can step away from the HDTV specifictions - can enable a true 1080p/60 service.

Then I'd archive that footage in my native editing codec to any form of data recorder - Blue Ray, Data Tape etc, but I would NOT encode it as MPEG-4.

[Jeff Kilgroe]

Although I wouldn't say interlaced video is superior to anything. Think 24p, 30p and 60p. (and 25p and 50p) But BluRay and friends can't do 50p or 60p at 1080 (do they allow 720p60 - I hope so), so they're not exactly future proofed. I just don't see why with HD they couldn't have made a nice clean break away from our analogue (ditch the 1000/1001 NTSC factor in frame rates) and standard def (interlacing) past.

Graeme

When I said 1080i was superior to 1080p30, I was just relaying that it handles fast motion better. It's the same number of pixels per second to the display, but (issues with interlacing aside), there is more temporal or perceivable information there to look at. Of course 60p is the way to go, if possible. As for BluRay, it does indeed support 1080p60 (and 720p60) and a multitude of other video resolutions and rates. I'm not sure what the current round of BluRay players support for formats in and display out, as all 1080p authored discs are just 1080p24... But I've attended a few Sony BluRay demonstrations / seminars and 1080p60 is definitely a format they intend to fully support, if it's not already present in the BDP-S1 player - it's there in the BluRay video spec. As are provisions for future resolutions beyond 1920x1080 as BluRay scales beyond two layers with a theoretical max of 12 layers. I've seen a 4-layer system demonstrated and that was this past July. I wouldn't be surprised if Sony is demonstrating 4-layer, maybe even 6 or 8 layer BD tech pblicly at NAB.

[Graeme Nattress]

Interesting Jeff. They'd either have to speed up the disc or go to much more efficient codecs to make 1080p60 work though?

Graeme

[Jeff Kilgroe]

But they didn't break from the 1000/1001 past, nor interlace either. Too much history and engineering to dig through there to worry about exactly why. Bottom line is HDTV broadcasting in the USA has two options - 720p/59.94 frames/ sec or 1080i/59.94 fields / sec. At a maximum of 22Mb/s in MPEG-2. Thats it. Anything else means a total rework of the DTV infrastucture.

But also in the USA, 1080p60 (as well as 1080p24 and 1080p30) is a fully specified format within ATSC broadcast standards. The real reason broadcasters mostly jumped to 1080i (and some to 720p) was due to initial digital conversion deadlines (which keep getting extended) and the hardware that was available. The only thing necessary for broadcasters to switch to 1080p is an upgrade to newer systems/hardware, which will be a gradual process. TV manufacturers also fought the 1080p shift for some reason that nobody can seem to figure out. They're all supporting it now though. My Samsung 71" I bought this last summer takes full 1080p60 input over HDMI (and 1080p 24/30 over component) and it's integrated ATSC tuner also supports 1080p24/30/60. So it's there... We're just waiting. For cable and satellite providers, it's going to be a long wait... We can't even get those guys to stop down-rezing the 1080i to 1280x1080. It's infuriating... I know that HD.Net absolutely hates that this happens to their content, but there's nothing they can really do about it since all the cable/sat providers do it to them.


What I'd be looking to create as an edited master would be 1080p/60, so that I can make a dub to 1080i/60 or 720p/60 for HDTV broadcasting and wait for when Blue-Ray, HD-DVD or internet or satellite based TV services - that can step away from the HDTV specifictions - can enable a true 1080p/60 service.

Yep.. If I was shooting content for anything not intended for theatrical/film distribution, I would go for 60p.

[Jeff Kilgroe]

Interesting Jeff. They'd either have to speed up the disc or go to much more efficient codecs to make 1080p60 work though?

Huh? BluRay discs can sustain 54Mbps, the video spec allows for 45Mbps + audio. Most current VC-1 and H.264 authored discs at 1080p24 have data rates between 9 and 22 Mbps -- yes, there's that much headroom and also why most current titles still don't look that great. Same for HD-DVD, which allows for a maximum rate of 37.5Mbps for video plus audio in their video playback spec, the disc tech itself handles a bit over 40Mbps.

What is needed is more layers and increased capacity. If you try to fit a 2.5 hour feature on a disc along with 40 minutes of extras, plus 50MB of ROM features all on a 50GB disc, you're looking at no more than a 28Mbps data rate, best case scenario. You might be able to push that to 30Mbps or more if you sacrifice quality or go SD res on the bonus features.

[GlennChan]

1080Psf30 versus 1080i60:

From what I gleaned from talking to Mike Prest (at a SMPTE meeting), perhaps the reason why 1080i60 (and 720p60) are favored over 1080Psf30 is:

-HD started being developed in circa 1980, with all the analog HD formats.

(paraphrasing loosely) Politically, the FCC and the Grand Alliance played a big part in this. Originally, the FCC was going to let the market decide which HD format would win out. The problem with that is that it usually ends up with a mishmash of proprietary formats. So, FCC later changed their mind and asked all the manufacturers and broadcasters to get together and develop a HD standard they can agree on. Hence the Grand Alliance.

Back in 1980s-ish, the CRT was still the dominant display technology. CRTs need to make at least 50 images a second, otherwise there is a visible and very annoying flicker. 50 isn't quite enough, but people in PAL countries get used to it. If you want to display 30 progressive frames a second, you'll run into this flicker problem. You can solve this by going with progressive segmented frame, which is basically using progressive frames in an quasi-interlaced system. Break the progressive frame in two fields, and show 60 'fields' a second. This avoids the flicker problem.

With analog systems, interlacing makes sense as CRTs don't have problems with interlacing (like plasma and LCD do). It's not a bad analog compression trick. With digital compression, interlacing is bad and less efficient than sending progressive frames. But anyways, the key point here is that CRTs were the dominant display technology when these standards were being developed. And for CRTs, interlacing presumably was a reasonable optimization of bandwidth.

*The ATSC Table 3 recognizes 18 different HD formats that may be useful/reasonable to implement. In that table, 1080p30 is recognized.

2- Still not sure what the reasoning is.

3- (personal opinion) If you look at ITU-R Rec. 709's adoption of the Rec. 709 luma coefficients, sometimes standards are adopted without actual testing. In hindsight, this was a bad decision as people implement the Rec. 601 luma coefficients instead, with penalties to color accuracy. However, ITU-R is a different standards body. As well, the Grand Alliance did have to show working prototypes.

The EBU has tested 1080i60 versus 720p60 versus 1080p60.

[Jared VanLeuven]

I'd always thought that the 720p/1080i broadcast cap was due to a bandwidth constraint over the existing freqs?

[GlennChan]

1080p60 has the following problems:
-No one is making 1080p60 cameras and other equipment yet (i.e. VTRs).
-Infrastructure-wise, it would ideally need to work with existing cabling. The higher bandwidth presents cabling issues... see
mms://www.windowsmedia.ryerson.ca/facultystreams/users/bfortner/smpte/sept2006/file1.wmv
(*I never listened to that presentation, but it should outline 1080p60 cabling issues)

Transmission-wise, trying to cram 1080p60 down the same bandwidth results in too many artifacts.
http://www.ebu.ch/en/technical/trev/trev_308-hdtv.pdf

[Jeff Kilgroe]

I'd always thought that the 720p/1080i broadcast cap was due to a bandwidth constraint over the existing freqs?

Well, 1080p30 and 1080i[60] take the same bandwidth... 1080p60 is obviously double that, but it's possible to send that over the same broadcast spectrum, where the real problem comes in (in the USA at least) is the FCC only allows each broadcast station a frequency window in which to operate and there's a lot of dead space in between. I don't know if these broacast "channel windows" are large enough for 1080p60 or not. I do know there were issues in the beginning of digital TV when a lot of broacasters wanted to push "datacasting". And some like FOX networks initially chose 480p as their digital standard so that they could reserve the remaining bandwidth for datacasting services... Stuff like up to the minute weather, news tickers, stock info, etc.. The FCC is fully aware that this may have to change at some point to account for future HD or other broadcast standards. I don't think they're willing to re-evaluate the situation now -- or at least not until the digital broadcast deadline has finally passed and they can shift their focus on the termination of the older analog system. At that point off in the future (originally slated for 2016, but has been delayed) they can re-distribute the older signal range, much of which returns to military and aviation use. I suppose in the future we will have to look to various forms of compression and other ways to squeeze the large amounts of data within broadcast windows. Otherwise all communications will have to shift over to cable or IP terrestrial networks and/or satellite.

Within the next 15 to 20 years, I think most of our television and other communications will predominantly be delivered by IP network. Either through a cable running to our homes or by wireless (WiMax, etc..). I think that traditional radio tower broadcasting as we know it today will eventually come to an end as sufficeint infrastructure based on other technologies allows for broad coverage and free access to essential localized news and services. Realistically, it could be just year or two before we see services like iTunes and Yahoo offering on-demand and subscription services in direct competition with cable and satellite providers.

[Greg Voevodsky]

Having seen Showscan which was 70mm at 60fps, I can say it did NOT have realistic blur eventhough wikpedia writes " SHOWSCAN is a cinema process developed by Douglas Trumbull. Like some other spectacular wide-screen processes, it utilizes 70mm film, but Showscan films and projects at a frame rate of 60 fps, 2.5 times as fast as standard cinema and twice as fast as video. It renders a picture that is not only extremely high in definition, but is dramatically smoother and more realistic in its rendering of motion."

WRONG, The rollercoaster that I saw showed the wooden cross section of the track tack sharp... In real life it is blured... and at 24fps or 30fps (more like video), it would look more realistic with real blur. Now, if they adjusted their shutter angle for more blur, it may have been realistic. The demo I saw was called hyper or super reality... not what wickpedia is doing rewriting history.

Everytime I have seen 60fps, the images like a video game until recently were not real. For example many times video games had cars driving at 75mph and then at 150mph and all they did was speed it up... the problem is at 150 which I have done on an empty two lane striped road in the middle of fenced farm country with no one around is that your vision shrinks to about 12 inch square... the dotted line becomes solid.. everything is blured except for a 1/2 a mile ahead, since you are dong 1/4 mile every 5 seconds... or 1 mile in 20 seconds... computer animators until a few years ago with some great motorcycle games that added blurr for everything above 120+ finally got what it feels like. So I do not believe 1080p 60fps would look right or realist... maybe hyper sharp and hyper realist - Showscan caused lots of people headaches at the time.. I can see why.

Also, when you encode to Blu-ray.. progressive does a better job. I have watched mpeg 2 destroy my DV footage with artifacts. I actually prefere 1080i for smoother motion to watch but for creating a DVD progressive will encode better and then if the TV interlaces 1080i, it will look great too.

[baro]

The EBU comparison is interesting and I think it is true.

Mark Schubin wrote in the 12 September 2006 Schubin's Memo (at IBC):

The European Broadcasting Union (EBU) offered an interesting compression demo. They stacked three identical large 1080p-capable HD displays. They shot material with a 1080p camera and downconverted it to 1080i and 720p. They fed all three types of signals into MPEG-4 AVC compression systems and fed the outputs of the systems into the displays. At extremely high data rates, all three looked good, and 1080p probably looked best. At low data rates, however -- even as high as 13 Mbps -- 720p looked better than 1080p; the latter stressed the compression too much. 720p looked better than 1080i, too.


http://www.digitaltelevision.com/mondaymemo/mlist/

[Graeme Nattress]

I think the EBU comparison makes a lot of sense. With digital transmission, interlace is very inneficient. I think 720p60 makes a lot of sense, however, lower frame rates should be transmitted as lower frame rates, not embedded into the 60 with a pseudo pulldown or whatever.

Graeme

[Andrew M.]

Looking on the cable bandwidth and bitrate per HD channel available now with 256-QAM of 38.81Mbps
we have enough speed there for one 1080p30 or 1080i60 with MPEG-2 and 4:1:1
But most cable companies use only 15-19Mbps as of now.
see link for QAM description:
http://www.scte.org/documents/pdf/ANSISCTE072000DVS031.pdf
page 21

I have such setup at home right now and the quality is far to low to call it HD, yet many people call it this way.

I have Sony PS-3 as a BR player and Yes, it has 50Mbps+ capability but I just watched two titles, CLICK and UltraViolet and the available bandwidth is not used at all. PS-3 has ability to show real time bitrate of the BR disk being plaid so you can check it all the time. CLICK BR disk was delivering in average 30 to 40 Mbps where UltraViolet only 15 to 20Mbps. Visual impression was exactly in the same proportions, UltraViolet looks like just the upres from DVD, CLICK looks bit better but I will not call it HD as I will not call HD current content delivery over the cable. Nice beginning though.
Anybody knows how CLICK was coded on the BR disk as oppose to UtraViolet?
Why such big difference in quality?

So no matter what numbers are shown on the specs of the equipment the visual impression of 2Mpixels material should be there as a final judge and the HD it is not there....yet.
RED is the only hope for the industry to deliver true HD to the media. 2Mpixels at 4:2:2 with very good codec clicking at 100Mbps is the minimum. Over the Air transmission could start using QAM modulation to double or quadruple the bitrate as cable TV does, but for now BR with double the throughput will do.
Is 100Mbps the magic number now for 2Mpixel material?
Maybe, but someone said once that we can not transmit more then 3Kbps in 3K of bandwith but this guy is dead now:-)

Andrew

[Jeff Kilgroe]

UltraViolet is a crap transfer. Nothing else to say there... Although, I question the quality of the source footage too -- it was shot with Sony F900U and heavly post-processed. I didn't think it looked that great as a theatrical release either.

Click was also shot with the F900 CineAlta and looks to be a better transfer than UltraViolet. It's encoded as MPEG2 and has a soft appearance. It's definitely better than DVD when viewed on a proper 1080p display, but doesn't really impress. IMO, I think MPEG2 BD titles are a waste of time and I'm puzzled as to why Sony and some of the other BluRay studios continue to release them.

I think the best HD transfers out there right now on BluRay (of what I've seen) are X-Men [3] and Ice Age - The Meltdown. The worst, well that's hard to say - there's lots of crappy BD transfers. But my vote goes to Crank that was just released. Good grief... I expect more out of standard DVD. The Brothers Grimm is a seriously poor transfer too...

HD-DVD has it's share of poor transfers too, but as a whole, I think HD-DVD titles look better. I credit this to the smaller capacity... HD-DVD authors had to jump right into using VC1 and newer codecs to make the releases fit on the disc as well as extras. With BluRay, studios have been lazy and continue to use MPEG2 for many releases and we're just getting DVD quality with about double the pixels. For HD-DVD though, my favorite transfer is the Corpse Bride. The black detail and color work is great... It gets my vote for best HD transfer on either BluRay or HD-DVD. Superman Returns is a pretty good transfer on all disc formats too.

[Graeme Nattress]

Click was Genesis though.

Sounds like the issue with new formats is as much the transfer as the format. You wonder why they don't employ some kind of quality control :-)

Graeme

[Jeff Kilgroe]

Genesis? Really? Got a link or something?

The Sony seminar I was at last July showing off BluRay, CineAlta and all the high-end Sony gear was showing tons of scenes from Click. The presenters were all saying it was shot with the CineAlta system. I didn't think Sony was using Genesis or any other high-end digicam other than their own.

As for quality control... It's the same thing we saw with DVD. There's a lot to be desired with many film transfers to DVD. I doubt we're going to get any better QC with the new HD formats. And to most consumers, all they care about is that it looks better than DVD, which doesn't take much when you connect a BD or HD-DVD player to a 1080p TV and push more pixels to it.

[Graeme Nattress]

Panavision were showing clips from Click in their trailer at CineGear as a demo for Genesis.

Problem is with a poor transfer, you're probably not even up to DVD standard for a lot of releases and putting more pixels on screen will be just more pixels of noise. Shame really, that people spend all this time to make a film look great, and someone comes along and puts out a crap looking disc of it.

Graeme

[Nick Shaw]

Genesis? Really? Got a link or something?


http://www.imdb.com/title/tt0389860/technical

[Rob Lohman]

You wonder why they don't employ some kind of quality control :-)

Costs money ;)

[Jeff Kilgroe]

Panavision were showing clips from Click in their trailer at CineGear as a demo for Genesis.

Interesting... Hmmm. And this could very well be the case (Click shot with Genesis). After all, we're talking about Sony who has no shame in hyping their own products, even if it's a big fat lie.


Problem is with a poor transfer, you're probably not even up to DVD standard for a lot of releases and putting more pixels on screen will be just more pixels of noise. Shame really, that people spend all this time to make a film look great, and someone comes along and puts out a crap looking disc of it.

Oh, I know it's a poor transfer... It's just really sad that with all the tools and resources the big studios have that they can't produce something better. One would think that Sony would be pushing the quality envelope with BluRay titles to try and out-shine HD-DVD. ...But no.

[Jeff Kilgroe]

http://www.imdb.com/title/tt0389860/technical

OK, that settles that one. Thanks, Nick.

So just more Sony hype then... I have no love for Sony, but mostly due to me spending a few years in game development. Between PSX and PS2, I've had all the Sony hype I can stand for one lifetime.

[Andrew M.]

Panavision were showing clips from Click in their trailer at CineGear as a demo for Genesis.

Problem is with a poor transfer, you're probably not even up to DVD standard for a lot of releases and putting more pixels on screen will be just more pixels of noise. Shame really, that people spend all this time to make a film look great, and someone comes along and puts out a crap looking disc of it.

Graeme

It is how I felt watching the CLICK and UltraViolet.
Watching it on the large screen in the cinema is acceptable, watching it off the BD is like watching DVD with 4 times the bitrate. Transferring from the film to the laser printer will do a better job.
Yet again if choices of codecs and transfer and transcoding method are poor the result is poor.
With HD there is more then pixels and bitrate it seems, all components in the chain have to be upgraded.
Good testing program/gizmo will give them good feedback, like we have for testing the PC systems or graphic card speed and performance. Sony Play Station-3 has only this bitrate display but this do not show the codec quality. Any idea how to rate the BD releases except watching them? (subjective)

Andrew

[tj williams]

So in the past we've been told to shoot on 24fps (actually 2397) for conversion to TV and possible film upconversion shows. Now you are saying 60P will look much better on tv (which I agree is true) so my question is how well can you derive 24P for a film out from this 60P master?

[Andrew M.]

Just double the frame rate and divide by 5

2 X 60 =120
120 / 5 = 24

So you can pick one frame every 5 frames or merge 5 frames in to one by special technic.
Photographers are taking 5 to 10 pictures of the same object and then combining it in to one in order to achieve best possible quality. I will not mention antiblur and color balance advantage coming out of combining 5 frames in to one.
http://www.cs.huji.ac.il/~peleg/papers/wacv00-blur.pdf
I think photoshop has these implemented in the latest version.

One more reason why RED has 120 frame rate capability.

[Graeme Nattress]

I'd personally like to see a framerate agnostic approach to TV and Cinema. Why should TV, not that it's just bytes, but locked to one frame rate? Surely, why convert PAL to NTSC when the TV could just start accepting the 25fps data (now that frames sizes are the same with HD), just like your projector does? Why shouldn't cinema projectors run at 50fps or 60fps if you want them too?

Graeme

[Anders Holck]

So in the past we've been told to shoot on 24fps (actually 2397) for conversion to TV and possible film upconversion shows. Now you are saying 60P will look much better on tv (which I agree is true) so my question is how well can you derive 24P for a film out from this 60P master?

No matter what resolution and well lit images, 60p will always look like "reality tv". If you want a cinematic feeling, 24p (or 25p) is the way to go, even for HD broadcast and tv conversion. In this case you'd want the strobing.
If you are shooting a golf tournament or a news show, 1080p60 would be preferable though.

[Graeme Nattress]

I think I'll take that as a challenge when we get our REDs to shoot 50fps or 60fps and make it still look totally cinematic without changing the frame rate :-)

Graeme

[Chris Kenny]

So in the past we've been told to shoot on 24fps (actually 2397) for conversion to TV and possible film upconversion shows. Now you are saying 60P will look much better on tv (which I agree is true) so my question is how well can you derive 24P for a film out from this 60P master?

Unlike 30p, 60p gives you enough frames that picking out 24 of them every second shouldn't introduce horribly noticeable motion artifacts. But note that you obviously can't shoot on a 1/48 shutter at 60p, so you're going to get less motion blur, creating a bit of a strobe effect if you remove frames. It might not be too noticeable if you shoot at 1/60, though, and I suspect clever software could use data from the removed frames to create fairly convincing motion blur for the 24p version.

A bigger issue is that while 60p will give smoother motion on television, it will also look more like TV news and less like what people think of as "cinema". This perception might change as technology creates more flexibility for acquisition and exhibition, though.

[Graeme Nattress]

However, if you shoot 60p, light it properly, and grade it nicely, it should look very nice and cinematic indeed. 90% of "cinematic" is lighting and grading.

60fps looks, traditionally, soap, because of harsh "video" lighting, video style image processing, edge enhancement and interlace. Avoid all of them and do 60fps progressive, then it will look very cinematic, but different, and perhaps better than 24fps.

Graeme

[Anders Holck]

Actually I have seen a beautifully lighted 75 fps 2k 35mm scan play back at 60 fps on a 720p plasma. While being very beautiful retaining all the organic properties of film, It just looked like really expensive high quality video to me.

[Graeme Nattress]

But perhaps that's because you know what to "look" for? It will be interesting to see what people can do with a high quality high fps source like RED?

Graeme

[Anders Holck]

Sure it will be interesting.
Say 24 fps was all about saving those remaining 36 fps in the celluloid budget, and 60 fps is the future...

BTV. Just saw a BluRay HD sampler from Sony on a 50" 1080p screen......never seen a better example of the horrors of edge enhancement and interlace twitter.
The "actors" (Models) were nice though :-)

[Graeme Nattress]

Interesting, enabling, and fun! BTW, I like the 24p film aesthetic too, but I also like the high quality high fps aesthetic too :-)

Um, if that's the Bluray sampler I've seen in local stores here in Ottawa, it's abysmal and whoever made it should be shot for crimes against quality!

Graeme

[Martin Jäger]

Hej, and hej graeme

just a quickie before i post my request: 24/25 fps will always look more cinematic than any significant higher framerates; it's more a psychological
fact than a visual matter. the lower the framerates, the more our brains place
the visuals into "past/happened" thus looking in a way more "poetic/historic"

ok;

now after i read your post where you (graeme) listed some codecs and
quality setings, i wanted to ask you if it'd be possible for you to make a quick
overview of how quality settings affect a specific codec - for example i didn't
know that photojpg turns to 4:4:4 when having it set to 100% - what about
90%? - how does the quality settings affect in animation-codec?

thanks for any hints/links/tables..

martin

[Graeme Nattress]

Martin, I prefer to think of it as a cultural phenonema, and a learned one at that, rather than a psycological one. There's nothing inherent in the frame rate that suggests "past / happened" other than growing up in a country that shows the vast majority of drama at 24fps.

Codec stuff is off topic for this thread - post over on Off Topic and we can pick up other there....

Graeme

[Martin Drew]

I sort of agree and disagree with you Graeme. I think that there is something significant psychologically going on with our perception of different frame rates, but that our interpretation of it is cultural and learned. To me high frame rates seem to look more real, low frame rates seem to look less real. It is not about an association with video it is something deeper than that. If you merely drop the interlace SD video instantly takes on a slightly dreamy feel even if the quality is pants.

Our association of low frame rates with cinema (ie high quality) and high frame rates with video (ie low quality) is purely cultural and we should definitely try to move forwards and get over that, but I suspect that there will always be a place for low frame rates to create a certain look, beyond trying to recreate film.

Martin

[Graeme Nattress]

Martin, you make some good points. And given that none of us can really divorce ourselves from our cultural upbringings, we'll never really know, other than we know what we like :-)

Personally, I'm looking for the day where we don't have a fixed fps per programme and can free form work with time as just another creative tool.

Graeme

[P Andersson]

keeps itching in my brain on this one

maybe graeme or another tech savvy person can explain

is it possible to allow for high framerates while allowing for slow shutterspeeds, of course it wasn't possible before as the film had to be transported away before the next picture started to get exposed, but now, perhaps one frame can start reading the sensor before the previous frame is finished, don't know at all if the technology allows for this

made a little illustration to explain the question

http://www.reduser.net/forum/uploaded/251_1169613716.jpg

[GlennChan]

I don't believe this increases sensitivity.

When you read a photosite, you take the charge out it. In between reads, the photosite is 'exposing'. The exposure time on each photosite is inversely proportional to the frame rate / read rate. Reading the photosites faster just decreases exposure time.

2- I think your scheme would have problems with motion. Assuming your shutter is fully open, all the photosites are always exposing but there will be a time offset between the photosites.

Related:
http://www.pixelink.com/support/oem/cameras/common_to_firewire/image_capture/rolling_shutter.htm

3- Another approach is to simply use a lot of light!

[P Andersson]

thanks glenn

if the information must leave the photosite on reading, this idea doesn't work

it was about adding blur to an image without having to use low frame rates

or having to make it a post thing

[Martin Jäger]

.. i understand your point about cultural habits of looking at somehing - but never the less you'd have kind of a strange, unconcious feelling if you'd watch the recent news on telly with 30fps instead of 60fields.. as Martin said,
higher framerates feel more real also because it's closer to our "internal" framerate of the brain. try to look at it like that: the longer the gaps between frames are the more information is missing - puting it to the extreme would mean you'd look at one take of a scene and you only show the first and the last frame... what happened in between? it'd be like looking at 2 photographs
- meaning - you have less gripp on what is "really" happening. - just my philosophy man :) i'm not claiming that i'm right..


martin

[Stuart English]

The good news is with RED we'll be able to experiment with the look of our images in a way not previously open to film or video cinematographers.

Having seen non-interlaced "smooth" video - albeit from a Varicam - I have a suspicion that 24fps ISN'T the key to a cinematic experience (heretic!)

Rather its the dynamic range, smooth and creamy tonal and color response, and lack of sharpened edges.

However, such a theory will soon be able to be tested.

[Greg Voevodsky]

As I said before seeing Showscan (google it) 70mm at 60 frames per second on a roller coaster, the front rung in front of the roller coaster was sharp just like its high shutterangle and not blurred having ridden the coaster in real life. It strobed. The psychology of film is how the eye retains a latient image so that we don't really see the strobe at 24 frames (actually 48 fps - half black) (being borderline) real life blur probably is around 30 frames.

I can tell you 60 frames unless you compensated for the shutter is way off. If it feels right in motion blur in real life and on digital cinema then it is... film is a bit slow. Video maybe a bit high... but I'm sure motion blur can be scientifically measured in the brain and on digital.