24p and quality

[Alex Huppenthal]

When Sony demos their 24p on the big screen are they actually representing that these images were created via a work flow using the camera's HDV compression?

Is it possible that 24p is being delivered uncompressed via the HDMI port into another workflow to create the material for the presentation?

I've been excited about new cameras before, and been dissappointed with the eventual quality. Frankly, Sony's magical non-HD resolution sensor producing quality HD video has me wondering.

I'd be very interested in this camera if it had a better than HDV compressed mpeg2 output, DVCProHD50 or higher rate output. The 25 mbit HDV format common denominator makes me skeptical. Compression technology has come along way since Mpeg2-ts.

I can see how 24p is less data than 30p or 60i, and how a much improved signal could be recovered from a given compression technology and data rate for a 24p capture.

This camera is a real puzzle to me. When the raw mpeg2-ts files hit the net, I'd love to check them out.

[Chris Hurd]

Originally Posted by Alex Huppenthal

When Sony demos their 24p on the big screen are they actually representing that these images were created via a work flow using the camera's HDV compression?

Yes. That was the whole point of the presentation.

Is it possible that 24p is being delivered uncompressed via the HDMI port into another workflow to create the material for the presentation?

No, because that would defeat the purpose of the presentation.

I've been excited about new cameras before, and been dissappointed with the eventual quality.

That's a shame. Quality is not determined by the camera. It's determined by the person operating the camera.

Frankly, Sony's magical non-HD resolution sensor producing quality HD video has me wondering.

It's no different from Canon's magical non-DV sensors in the old XL1 and XL1S, or Panasonic's magical non-HD sensors in the HVX200. Frankly, the people who regularly use those cameras on a consistent basis -- that is, those who know their gear -- find the images they're getting to be perfectly acceptable for what they're doing, including professional broadcast and theatrically released filmmaking. If there were some "problem" in this regard, then the cameras wouldn't sell, but the fact is they've been selling like crazy.

I'd be very interested in this camera if it had a better than HDV compressed mpeg2 output, DVCProHD50 or higher rate output.

Well, if it had a better than HDV compressed mpeg2 output, then it wouldn't be an HDV camcorder anymore. There is no such thing as DVCProHD50 -- there are instead only the standard definition digital video formats of DVCPRO and DVCPRO 50, and the high definition format of DVCPRO HD -- but if this camera had one of those particular formats, then it certainly wouldn't be a Sony anymore. It would be Panasonic! Because those formats are strictly Panasonic teritory. Sony offers other HD formats with higher rate outputs, but that's getting you into the territory of HDCAM (on tape) and XDCAM HD (on disc), but they are of course at a significantly higher price bracket than HDV. Point is, the choices are there if you want them.

The 25 mbit HDV format common denominator makes me skeptical. Compression technology has come along way since Mpeg2-ts.

It's crucial to understand that compression does *not* affect quality. One look at high definition H.264 video on a thirty-foot screen projected in the 8mbps AVC codec will convince you of that (seeing is believing, after all). Try not to get hung up on numbers such as the bit rate. What you really need to be considering is how many other people in different types of markets (from hobbyist to Hollywood) are successfully adopting HDV. Look around at the various user reports, and even more importantly, try it out for yourself.

I can see how 24p is less data than 30p or 60i, and how a much improved signal could be recovered from a given compression technology and data rate for a 24p capture.

Entirely the wrong way to look at it. 24p is all about the motion signature of that particular frame rate; it's all about a certain aesthetic character of motion. Choosing to shoot in 24p is a decision which is all about the *frame rate* and you're doing yourself a disservice to consider it otherwise.

This camera is a real puzzle to me. When the raw mpeg2-ts files hit the net, I'd love to check them out.

I think the "puzzle" has less to do with the camera itself, and more to do with some fundamental misconceptions about the HDV format arising from a lack of direct experience with it. Checking out video clips from the net is fine -- that is, as long as you're viewing them on an actual HDTV and not a computer monitor -- but ultimately the best thing for you to do, if you're really serious about this, is to get one into your own hands and try it out for yourself. Hope this helps,

[Heath McKnight]

Great answers, Chris. Even after 2-3 years, HDV is a mystery to some people. As everyone knows, I work about 60% with HDV (Z1 or FX1), the other 40% is with the DVX100a. I've never had issues and am quite happy with the results.

For Z1 footage (50i) converted to 24p, check out:

www.myspace.com/904am

Or do a search here and find native m2t V1 footage from Douglas Spotted Eagle. He links to his site.

heath

[Heath McKnight]

Spot's video:

http://www.dvinfo.net/conf/showthread.php?t=75982

hwm

[Philip Williams]

Alex, put very simply HDV has the following attributes:
1. It is capable of video quality that will keep paying clients happy.
2. It is capable of video quality that will not distract users from a narrative project.

I've only had my XH A1 for a couple of days and I already know one thing for sure: its better than I am. I am absolutely 100% convinced that viewers will notice bad lighting, bad composition and bad acting LONG before they notice any HDV shortcomings.

Truthfully, I think someone spending 4-6K on a nice 3 chip cam these days can hardly go wrong. XH A1, V1, HVX, HD100, whatever - as long as you pick the right style of camera for your needs (lens configuration, frame rates, etc..) you will have a tool capable of turning your creativity into amazing imagery.

[Alex Huppenthal]

Originally Posted by Philip Williams

Alex, put very simply HDV has the following attributes:
1. It is capable of video quality that will keep paying clients happy.
2. It is capable of video quality that will not distract users from a narrative project.

I've only had my XH A1 for a couple of days and I already know one thing for sure: its better than I am. I am absolutely 100% convinced that viewers will notice bad lighting, bad composition and bad acting LONG before they notice any HDV shortcomings.

Truthfully, I think someone spending 4-6K on a nice 3 chip cam these days can hardly go wrong. XH A1, V1, HVX, HD100, whatever - as long as you pick the right style of camera for your needs (lens configuration, frame rates, etc..) you will have a tool capable of turning your creativity into amazing imagery.

Thanks for that. I agree those 3 chip HD cameras are good. Practical, useful tools. And all worth buying. The job defines the camera. I could take an Elura 100 out and achieve a nice music video. Judging by the thousands of videos produced recently, web cams work too - see youtube. If you are doing it for free. well, free is a type of work, isn't it?

I have a small Sony HVR-A1U. I owned the JVC HD 10U for years and produced various documentary videos. We travel to mountain tops and document Radio installations, mostly industrial, but the views are incredible. We're moving on to broader documentary films.

I have seen H.264 compressed video and its often great. 2 Pass compressors can do amazing things. Limit artifacts, deal with frame content changes and adapt. I'm familiar with all compression technology of recent history. Bit rates, compression technology have everything to do with quality product delivery. I don't know any idiot proof, one touch production system that will take the output from an NLE and produce an outstanding product.

But I'm not writing to fence about which compressor does what job better. Workflow is a hard won process. Suffice to say H.264 and Mpeg4 beat Mpeg2 in many many ways. I can produce a 1 hour HD video on a standard DVD in H.264 format that rivals any other HD delivery method for image quality. I would like to be able to do it easily from a $5,000 camera. :-) Panasonic steps up nicely. I'm about to aquire 2.

My point was simply that I'm a bit skeptical, and Chris, not about the quality I can personally produce. The JVC HD 10U at 720p produced some amazingly wonderful images. I'm very happy with that camera. I've found the little Sony A1U to be less of a camera, IMHO. That's the root of some skepticism. In part due to (i) part of 1080i.

Recenly I've seen a couple of Z1's clearly outdone by the Panasonic DVCPRO HD video encoder.. :-) I'd been on the fence between Sony and Panasonic, but now that HVX200 are selling for $4200, that's my camera to beat.

Neither of my 2 HD cams are 3 chip cameras, and I have yet to spend much time with a 3 chip camera other than the HVX-200. Its quality at 720p is very very good, and its easy to produce results that meet my opinion of quality.

I agree that seeing is believing. With a frame capture out of the digital image stream, and some hands on with the source, I can tell if it'll be a good fit for me, and that's something I'm looking forward to.

In the mean time, I'll stick with my view that the proper camera design will deliver 1080x1920 frames at either ~30 or ~24 FPS for me, at minimum, directly out of the imager. I'll be thrilled when mini-RED is here.. :-)

Recall the Sony battle to produce the famous Mini-disk recorder compressor, (which records raw pcm audio now, by the way, and makes it a great little audio recorder). Perceptual encoding is a battle that can go on forever, or, until you are getting 30 fps of full images, not the differences between images, or the myriad of techniques used to approximate film.

The fact that some people can't tell there are parts of an image missing or parts of the audio missing, doesn't mean they are there, they are missing!

From my purist view, you expect the video technology to achieve parity with still image tech. I can drop 11 frames of RAW 5000x3300 image into a buffer, and 22 Jpeg frames at that resolution. Okay its at 4 fps, but we're talking video 1080x1920 here, not 5K. (talking 1dsm2, a still camera i own) dropping 16 unique megapixels into a buffer at 4 fps is pretty impressive. Do the math. that's 16 * 4 = 64 megapixels /sec. That's 2 year old technology. I want 2 megapixels of unique images at 24 fps = 48 megapixels / sec.. The quality can be as bad as my 1DsMark2 camera. That's acceptable. :-)

Recenly I watched a timelapse from a 1Ds turned into video. Purely amazing quality. My point is, push the technology, innovate, damn it. Be creative.

I'm a fan of Sony's. I own a bunch of their equipment. I'm looking forward to this new system, for the reason I said. I'm skeptical. :-)

There is no reason anyone else should be. Clients, and pay makes the world go round. If the output suits the task, you have a winner.

I expect real 1080p now. I expect real 4K in 2 years. Sorry if that seems pushy.

My post is simply my opinion, and desire.

[Douglas Spotted Eagle]

Originally Posted by Alex Huppenthal

I agree that seeing is believing. With a frame capture out of the digital image stream, and some hands on with the source, I can tell if it'll be a good fit for me, and that's something I'm looking forward to.

If by "frame capture" you're referring to stills captured from the source, I'd challenge that you can determine much from those images.

Originally Posted by Alex Huppenthal

From my purist view, you expect the video technology to achieve parity with still image tech. I can drop 11 frames of RAW 5000x3300 image into a buffer, and 22 Jpeg frames at that resolution. Okay its at 4 fps, but we're talking video 1080x1920 here, not 5K. (talking 1dsm2, a still camera i own) dropping 16 unique megapixels into a buffer at 4 fps is pretty impressive. Do the math. that's 16 * 4 = 64 megapixels /sec. That's 2 year old technology. I want 2 megapixels of unique images at 24 fps = 48 megapixels / sec.. The quality can be as bad as my 1DsMark2 camera. That's acceptable. :-)

I expect real 1080p now. I expect real 4K in 2 years. Sorry if that seems pushy.

When video camcorders sell in quantitiy equal to that of mid-level still cameras, ie; Canon Rebel XT, then you'll see parity in cost/performance. Until then (which is likely on the 12th of never), you'll not see parity in the ranks. Cost-wise, it's simply impossible for manufacturers to provide with the sales-generated development allocation $$ that manufacturers use.
As far as 1080p, it's a bandwidth issue. Nothing affordable shoots it well because the bandwidth isn't there for 1080p60. Do the math. Either compromise the pixel content (DVCPro method), compromise the frame content (MPEG method), compromise the color space (both DVCPro and MPEG method) or create new technology that allows for compromises all around (AVCHD method) if you want to stay affordable. Or, find fatter, cheaper pipes to allow for encoding and storage.
At least we're no longer encumbered on all levels by the "train tracks" of storage methods. The rails can be wider or more narrow today, thanks to new technology, but widening the rails calls for a much bigger engine (CPU). CPU speed must also be part of the equation when considering all this. Even if we had an affordable 4K camera in 2 years, where is the CPU that can manage it as a compressed stream, in an affordable package?

[Alex Huppenthal]

Originally Posted by Douglas Spotted Eagle

If by "frame capture" you're referring to stills captured from the source, I'd challenge that you can determine much from those images.




When video camcorders sell in quantitiy equal to that of mid-level still cameras, ie; Canon Rebel XT, then you'll see parity in cost/performance. Until then (which is likely on the 12th of never), you'll not see parity in the ranks. Cost-wise, it's simply impossible for manufacturers to provide with the sales-generated development allocation $$ that manufacturers use.
As far as 1080p, it's a bandwidth issue. Nothing affordable shoots it well because the bandwidth isn't there for 1080p60. Do the math. Either compromise the pixel content (DVCPro method), compromise the frame content (MPEG method), compromise the color space (both DVCPro and MPEG method) or create new technology that allows for compromises all around (AVCHD method) if you want to stay affordable. Or, find fatter, cheaper pipes to allow for encoding and storage.
At least we're no longer encumbered on all levels by the "train tracks" of storage methods. The rails can be wider or more narrow today, thanks to new technology, but widening the rails calls for a much bigger engine (CPU). CPU speed must also be part of the equation when considering all this. Even if we had an affordable 4K camera in 2 years, where is the CPU that can manage it as a compressed stream, in an affordable package?

Good point, 1DsM2 isn't a mid level camera, but I agree with you, the volumes are not there. Perhaps we're in catch 22 space at present. Not enough volume means one can't build a camera that will develop an extremely large following. Canon has the resources, as does Sony to make the corporate commitment required to *make* the market and not follow it.

Back to the new Sony. As I see it 1080/24p is magically pulled down, up, so that 60i HDV is created. I understand their site says it maintains compatibility with existing systems. That serves the market.

Is that 24p to 60i conversion even possible? Does anyone have a feeling for how this process will work in real life. I know I'm asking for speculation here.

and 24p @ 1080 does not require as much headroom as 1080/30p does, so, does that explain the 60i demos being soft? I can see the magic of using the extra 6 frames per second of headroom 24p offers to enhance what's there, I just can't get my brain around how you don't destroy that going to 60i on the tape, with my somewhat limited understanding of Sony's pulldown strategy.

:-) Its a mad man's heaven this video tech stuff. I know I'm not the only one who'd like to see (NNi) dissappear, the words pulldowns and telecine eliminated. The world is moving smartly toward digital delivery, and with some luck (I have my fingers crossed) cameras like Sony's V1 will move us closer to that.

I'll be happy to read the 'Functions" and "Design" section of their website
http://bssc.sel.sony.com/Broadcastan...V1U/index.html

when it gets completed. As of today "Coming Soon" is still there.

I dig the tapeless accessory. Also if they move toward some recording format that's not HDV to that device, man, we're talking a breakthrough camera.

Its hard to stop speculating about the quality. I have my finger on the 'buy-it' trigger.

Anyone know where you can see one in the Pacific NW? Portland/Seattle?

[Barry Green]

Originally Posted by Alex Huppenthal

Is that 24p to 60i conversion even possible? Does anyone have a feeling for how this process will work in real life.

It's done every day. Any Hollywood movie broadcast on television has gone through that process. Any Hollywood DVD you've ever watched on a CRT television has gone through the process. Any film you've ever seen transferred to video has gone through that process. It's simple 2:3 pulldown, where two frames of video get spread across five video fields.

[Heath McKnight]

Most TV's specs include info on a 2:3 pulldown, if you look at it.

heath

[Alex Huppenthal]

So, I'm still trying to get my mind around the Sony approach to 24p -> 60i. If they do it poorly, the result will be no advantage to 24p over 60i in terms of quality possible (read this as compression+bandwidth) on the HDV tape (or disk). hmm.. the repeat frame-repeat frame stuff on tape is a waste of time/space, as I understand it. True?

[Douglas Spotted Eagle]

No, it's not a waste of time/space. If you live in a world where you control everything, have no outside responsibility, have no legacy requirements, and no one to answer to but yourself, you could create a proprietary workflow for acquisition, editing, delivery. Until then, there are always compromises that can work best for all parties involved. That's essentially what pulldown is; a compromise to meet world needs, film needs, and delivery to the masses.

[Heath McKnight]

Alex,

Do you use the DVX100, a or b series? How about the XL2, or the Canon HDV offerings? How about nearly all 24p video cameras of varying formats? Pretty much all of them use 24p in a 60i stream.

heath

Originally Posted by Alex Huppenthal

So, I'm still trying to get my mind around the Sony approach to 24p -> 60i. If they do it poorly, the result will be no advantage to 24p over 60i in terms of quality possible (read this as compression+bandwidth) on the HDV tape (or disk). hmm.. the repeat frame-repeat frame stuff on tape is a waste of time/space, as I understand it. True?

[Alex Huppenthal]

I use the HVX200, A1, and HD10, if that's important.

Are you saying all the cameras you mentioned turn 1080/24p into 1080/60i in their workflow? They aren't restricted to HDV, however. are they?

Perhaps I need more info. HDV is a 1080/60i format. Camera capabilities are throttled a bit by how the video leaves the camera. If it exits at 1080/24p and is recorded @ 1080/60i well, I'm learning this is just how it is with HDV stadnard recording. I guess the rev to HDV recording is AVCHD.

I'll go study the AVCHD offerings.

[Heath McKnight]

Alex,

I'd say 90% of video cameras (DV, SD, HDV, DVCPro HD, HDCAM, etc.) put the 24p stream into a 60i signal for delivery. You remove the pulldown to get 24p out of it. It retains the look of 24p, though. I believe the only setting on the HVX200 that is native 24p is 720p 24pn. The others utilize the pulldown.

The pulldown is new to HDV, as is 24p (original spec called for 720p30/25 and 1080i60/50, along with 720p60, I believe). If you want to know more, check out the second edition of VASST's HDV: What You NEED to Know, which I helped write:

http://www.vasst.com/product.aspx?id...0-0ad68139e940

Hope this makes it more clear.

heath