AVCHD vs AVCintra out of the AF 100 article by Crewstv [Archive] - DVXuser.com -- The online community for filmmaking

PDA

View Full Version : AVCHD vs AVCintra out of the AF 100 article by Crewstv

Everts

10-21-2010, 06:54 PM

http://crews.tv/blog/2010/10/22/af100-images-pushed-to-the-limit/

Homunculus

10-21-2010, 07:27 PM

BowiesInSpace

10-21-2010, 07:44 PM

as i said in other thread, very encouraging. they are saying in - cam native acvhd is virtually indistinguishable from AVC intra external recorded footage. could this be because there IS no difference due to the port still being 8bit even to external recorder?

Even if the output is 8-bit there's still no restrictions on recording to 4:2:2 or 100mbps on an external recorder.

I'd say Jan was right. So many people were dissing the native codec, but we should wait to see if we're pleasantly surprised by it. I sure am.

Definitely not spending a few grand for a nanoflash asap. Better to wait until my cell phone shoots 3K, then I'll really need to up my camcorder codec.

Michael Olsen

10-21-2010, 08:04 PM

Philip Goetz

10-21-2010, 08:08 PM

I am thinking that AVCHD 25mb/sec recording to SDHC cards in the camera is just a way to get off the ground and rolling to get recording. Using an external recorder will be the key to high resolution digital cinema.

Michael Olsen

10-21-2010, 08:09 PM

It would be interesting to see a similar test from the footage of the Alfa Romeos - something with more motion that pushes the codec a bit.

Jaime Valles

10-21-2010, 08:24 PM

Very encouraging. More tests are in order, but I'd say this is making me reconsider the need for a Ninja or a Ki Pro Mini for serious production work.

Homunculus

10-21-2010, 08:28 PM

I must say I prefer the look of the AVC-I. The difference is very subtle, and strikes me most in the bokeh. Seems like the AVCHD footage is a slightly more "square" in its rendition of detail. Of course, this is just looking at 1920x1080 JPEGs, sooo...

This is definitely a head and shoulders above DSLR h264. :beer:
wow you seriously can see a difference? You must have bionic eyes. are you a replicant? lol I flipped between the 2 images instantaneously back and forth for 2 min and could not see an iota of difference. am i blind??

TheReverend

10-21-2010, 08:28 PM

Matthew B

10-21-2010, 08:31 PM

Very encouraging. More tests are in order, but I'd say this is making me reconsider the need for a Ninja or a Ki Pro Mini for serious production work.

For serious production work i'd still consider an external recorder. They said the codec held up in basic colour correction, but i think where the codec will fall apart will be during much heavier correction, not to mention if you wanted to do some serious greenscreen work.

Homunculus

10-21-2010, 08:31 PM

When seeing this after the "other" thread... I told you so. :D

But seriously this is good news. Hopefully people will realize that they didn't just ignore what we want. They gave a great codec inside the cam, and an even better output, at a great price. Well done Panny and Jan!

hmm in 'other thread' were you the proponent of the theory that no matter how much mbps you increase the bitrate of the codec it will always stay the same in avvcam ? i.e. quality will not appreciably increase at all? why is this?

wouldn't this mean that ki-pro, ninja, whatever are all literally useless for af100?

Matthew B

10-21-2010, 08:34 PM

wow you seriously can see a difference? You must have bionic eyes. are you a replicant? lol I flipped between the 2 images instantaneously back and forth for 2 min and could not see an iota of difference. am i blind??

I can see a difference aswell, only slight though. To me it seems that the AVC-I was a bit more contrasty

Matthew B

10-21-2010, 08:39 PM

Chris Messineo

10-21-2010, 08:41 PM

Great article. I don't always understand all the bits and bytes of every argument on DVXuser, but I can appreciate what my eyes see and the AVCHD footage so far look beautiful to me.

I'm probably in the minority on this, but I'm thrilled to have a codec that can deliver quality results without filling my hard drive every time I shoot.

I'm so excited that I preordered and I can't wait for December to get here.

dcloud

10-21-2010, 08:44 PM

theo nly difference i see is color rendition with avcintra having more color gradients

Steve Kahn

10-21-2010, 08:54 PM

In these AVCHD does look good. To be very picky one can a slight bit more noise in the blacks in Frame-2. To my eyes both Frame-1 look virtually identical

Joe Lawry

10-21-2010, 08:59 PM

They codecs rendition of the scenes we shot are so close that everyone i've shown in my edit suite, cannot tell them apart. DPs, Camera Operators & Editors.. they all were amazed.

I was the one who extracted those frame grabs for that article, they were originally full res tiff files however i think hosting them was an issue.

I've put the JPEG and TIFF side by side and they are 99% identical.

AVCHD FTW.

TheReverend

10-21-2010, 09:01 PM

hmm in 'other thread' were you the proponent of the theory that no matter how much mbps you increase the bitrate of the codec it will always stay the same in avvcam ? i.e. quality will not appreciably increase at all? why is this?

wouldn't this mean that ki-pro, ninja, whatever are all literally useless for af100?
Ummm, that's not really what I said. I stated many times that a proper implementation of AVC at 24mbps will be more than enough for most scenarios. This report supports that. I did state that more bitrate won't necessarily help get a better image. Again this is backed up by this report because only when adding 4:2:2 and I frames only at 100Mbps do we see improvement, and marginal at that.

The end story is 24Mbps with I B and P frames is a great balance for quality and data rate. And that's good news for anyone looking at this cam.

Justyn

10-21-2010, 09:11 PM

Ive been shooting with the 150 for almost two years and the codec can really handle a lot of CC and even on the greenscreen it's been virtually identical to what I was able to get off the HVX, with one thing to note, the HVX was a bit more forgiving about off lighting and shadows and less than ideal keys, but if you do your job, then this is NOT an issue for the codec. I have shot a few hours of greenscreen with the 150 with great success...

This is super good news all in all cause I really didn't want to take away the simplicity and complicate things. This was supposed to get us away from the kitchen sink setup of the DSLRs, and I for one and thankful for that. My only decision is whether or not to sell my 150....

Michael Olsen

10-21-2010, 09:11 PM

wow you seriously can see a difference?

Yes. As I said, it's very subtle. I'm looking at it on a 21" 1080p monitor. If I'm about four feet away, I can't distinguish between the two. But the AVC-I footage is somehow a little smoother looking to me. I was able to distinguish it without knowing which image was which first. In the real world, when looking at moving footage and not a big compressed JPEG, I don't think even that subtle difference would really appear. I would, however, be very interested to see how the codecs affect noise when playing back - something that can't be seen in a JPEG.

Frankly, this test is quite favorable to the AVCHD codec. I think the real test will come with lots of detail in the scene - a flyover of a forest or a pan of lace - and/or motion.

You must have bionic eyes. are you a replicant?

Tortoise, what's that?

Duke M.

10-21-2010, 09:42 PM

hyphygreek

10-21-2010, 10:33 PM

This looks great... I'm looking at it on my 30" monitor and there's not really a difference at all.

-JJ Labritakis

Lyris

10-21-2010, 10:34 PM

I'm not surprised at this finding. Unless you are recording some sort of strobing laser-light show, then AVCHD appears to be a smart compromise.

Granted those are static scenes, but even still it's looking great. I would expect the differences to be more apparent with more difficult footage (water, sparks etc.) but really, what do you expect?

Edit: nice to see some full-frame images from the camera, too, to finally judge the resolution. I look forward to seeing more images from different lenses, and obviously from the final camera as well.

Lyris

10-21-2010, 10:43 PM

To me it seems that the AVC-I was a bit more contrasty
It's not. The tonal distribution is identical on both of those images. I'd be worried if one of those expensive professional recorders was skewing the levels.

trez

10-21-2010, 11:14 PM

It is especially impressive that people can spot the difference in the bokeh, where all details are blurred (?!)...
Contrast is sightly different in both versions, agree. This is not enough to say one is better than the other, though.

John Caballero

10-21-2010, 11:22 PM

Originally Posted by Michael Olsen
I must say I prefer the look of the AVC-I. The difference is very subtle, and strikes me most in the bokeh. Seems like the AVCHD footage is a slightly more "square" in its rendition of detail. Of course, this is just looking at 1920x1080 JPEGs, sooo...

This is definitely a head and shoulders above DSLR h264.

People see whatever they want to see.

This looks great... I'm looking at it on my 30" monitor and there's not really a difference at all.

And there won't be.

Matthew B

10-21-2010, 11:23 PM

It is especially impressive that people can spot the difference in the bokeh, where all details are blurred (?!)...
I agree contrast is sightly different in both versions, agree. This is not enough to say one is better than the other, though.

That is all i meant, is that it looks SLIGHTLY different to me, not saying that one is better than the other.

DylanReeve

10-22-2010, 12:24 AM

It would be interesting to see a similar test from the footage of the Alfa Romeos - something with more motion that pushes the codec a bit.

Indeed. The footage available for us on this was pretty optimal really, but it's good because it provides us a pretty decent baseline. I'd also love to see similar tests on the Alfa footage, it's pretty much the extreme of real-world usage (unless you're racing cars through a forrest).

For serious production work i'd still consider an external recorder. They said the codec held up in basic colour correction, but i think where the codec will fall apart will be during much heavier correction, not to mention if you wanted to do some serious greenscreen work.

I actualy just went crazy with the curves and hue wheels in Media Composer's colour corrector - it didn't break anymore than I'd expect from any footage... There's only so far you can push something, but overall pushing the footage didn't highlight any compression issues as we see with DSLR footage quite often.

Ideally I'd want a better codec for greenscreen work, as 4:2:0 is always going to make that more difficult. But that said, I wouldn't be too freaked out pulling a key from this. I've done greenscreen on a 5D, and I'm pretty certain the AF100 would do a fair bit better than that.

Yes. As I said, it's very subtle. I'm looking at it on a 21" 1080p monitor. If I'm about four feet away, I can't distinguish between the two. But the AVC-I footage is somehow a little smoother looking to me. I was able to distinguish it without knowing which image was which first. In the real world, when looking at moving footage and not a big compressed JPEG, I don't think even that subtle difference would really appear. I would, however, be very interested to see how the codecs affect noise when playing back - something that can't be seen in a JPEG.

A still frame is a little unfair because it isn't really a still image - noise, compression or glitches that appear in one frame might not be the same in the next frame. The JPEG's are pretty high-quality, we can probably also provide the full-res TIFF files. I'll check on that. There's two more frames that we've not showed in the article there.

Frankly, this test is quite favorable to the AVCHD codec. I think the real test will come with lots of detail in the scene - a flyover of a forest or a pan of lace - and/or motion.

Yeah, it wasn't really pushing the codec too much, but I actually quite like that as it lets me see what the baseline codec can deliver. I know then that I'm not likely to see worse results except if more exceptional circumstances. It would be good if we had a few high-motion samples too, but the guys just didn't have time to fit it all in sadly.

as i said in other thread, very encouraging. they are saying in - cam native acvhd is virtually indistinguishable from AVC intra external recorded footage. could this be because there IS no difference due to the port still being 8bit even to external recorder?

The 8-bit thing isn't major - it will make a difference obviously, but the image is optimised for it. The big differences will be the 4:2:0 colour onboard, and the Long GOP interframe compression. The blowup in the article demonstrates the impact that 4:2:0 will make, and the interframe will start to become relevant as the detail and motion increases. If you're shooting high detail, or a lot of motion then an off-board 4:2:2 recorder will make itself much more valuable.

I must say I prefer the look of the AVC-I. The difference is very subtle, and strikes me most in the bokeh. Seems like the AVCHD footage is a slightly more "square" in its rendition of detail. Of course, this is just looking at 1920x1080 JPEGs, sooo...

This is definitely a head and shoulders above DSLR h264. :beer:

The 4:2:0 colour sampling, combined with the lower-bitrate DCT seems to create slightly more obvious blocking, but if anything that should be less obvious in areas of soft focus.

Contrast is sightly different in both versions, agree. This is not enough to say one is better than the other, though.

I can assure you there is no difference in contrast in the source images. If you're seeing some difference then it must be a quirk of the JPEGs or something.

Lyris

10-22-2010, 12:27 AM

For a more scientific approach, the SSIM of the AVCHD version of the closeup image compared to the AVC-Intra is 0.9700.

An SSIM result of "1.0000" would be an exact match. 0.9700 is excellent, in fact better than I was hoping for.

trez

10-22-2010, 12:52 AM

[QUOTE=DylanReeve;2137577I can assure you there is no difference in contrast in the source images. If you're seeing some difference then it must be a quirk of the JPEGs or something.[/QUOTE]

Looking at the stills again carefully... And I think you're right. It was probably just different viewing angle (was looking at them on my laptop LCD).

DylanReeve

10-22-2010, 12:57 AM

Hmm. My SSIM implementation must be a little wonky, because I get 0.99733... Should look at that (a program I wrote myself based on Chris Lomont's code (http://www.lomont.org/Software/))

Anyway, I actually get a higher result (0.99772) using the uncompressed TIFF grabs of that same stil, so at least some of that is JPEG compression.

J Davis

10-22-2010, 01:12 AM

great tests and encouraging results ... can't wait to shoot one

Michael Olsen

10-22-2010, 01:24 AM

People see whatever they want to see.

When I viewed the images I made sure to do so under blind circumstances - I didn't know which was which. The one I favored turned out to the AVC-I both times.

Again, I doubt this really means anything practically. What I can perceive is a very minor textural difference that disappears with a little bit of distance and, frankly, would probably be invisible when the footage is in motion. After all, this isn't a still photography camera! :beer:

Yves Simard

10-22-2010, 04:11 AM

People see whatever they want to see.

This is especially true of the DSLR - myself included.

Barry_Green

10-22-2010, 04:39 AM

I would expect that in extreme pixel-peeping the AVC-I should hold up better, it should show more levels, and it should do better for greenscreening, and it should look very slightly sharper. I mean, it'd better -- it's 4x-5x as much bandwidth, etc. Quite impressed with the AVCHD showing here, because IINM, the camera crews.tv was using was the same "generation" as the one I used, and they specifically told us *not* to use the onboard codec for anything serious yet, as they were still improving it! So it's possible that the final version will actually look even better than this.

Yak

10-22-2010, 05:01 AM

I love this camera , but in comparing AVCcam and AVCIntra shots of a relatively static sean one should not see much difference, the main difference between the codec is one is all I frames the other uses I,P & B frames. When there is little or no movement the lack of all I frames should not show up on AVCcam only when the codec is stressed with quick changes and lots of detail will the difference be come evident.

Everts

10-22-2010, 05:03 AM

I would expect that in extreme pixel-peeping the AVC-I should hold up better, .....

They did that at 600%,right ?

Homunculus

10-22-2010, 07:41 AM

When I viewed the images I made sure to do so under blind circumstances - I didn't know which was which. The one I favored turned out to the AVC-I both times.

Again, I doubt this really means anything practically. What I can perceive is a very minor textural difference that disappears with a little bit of distance and, frankly, would probably be invisible when the footage is in motion. After all, this isn't a still photography camera! :beer:

man someone give this guy a Turing test already

larryfine451

10-22-2010, 08:35 AM

Well I am definitely looking forward to hands on testing with this cam ...and its onboard will most lkely one of the best implemented versions of AVCHD to date . and at ~5k a good offering

But I am a little dissapointed but not suprised at lack of major improvement by outboarding capture .
One can only imagine what kind of images could be produced by 10 or 12 bit output...

Perhaps we could have another model at a higher price which could provide that option...8-10k

f64manray

10-22-2010, 09:23 AM

Jan has always said that the AVCHD as implemented in the AF100 was the little brother to AVC-Intra. Who knew they would be twin brothers?

I don't see at this point how someone justifies spending $2-4K on an external recorder after looking at these results unless everything just flies apart when your actor starts walking. Somehow, I don't think that will be the case.

Thomas Church

10-22-2010, 09:55 AM

It seems I'm skipping an external recorder altogether, unless I actually shoot a feature with the AF100.

For all smaller jobs I'm gonna go with AVCHD alright.

Mike Sertic

10-22-2010, 10:15 AM

I don't see at this point how someone justifies spending $2-4K on an external recorder after looking at these results unless everything just flies apart when your actor starts walking. Somehow, I don't think that will be the case.

I have a HMC40 and have used it in combination with a nanoflash at 140mb/s I-frame shooting surfing footage, which is very challenging to long-gop compression due to lots of random water motion and changing textures.

I realize that the images to be encoded from the two cameras are very different due to the sensor size and resultant depth of field, noise levels, etc., but I assume the AVCHD implementation is going to be somewhat similar as they are both Panasonic AVCAM products.

That said, if the AVCHD implementation on the AF100 is anything like the HMC40, static shots between nanoflash files and the AVCAM will look quite similar except for the difference between 4:2:0 and 4:2:2 (which I really only can notice at edges of high color contrast in static shots), and lots of motion or really-high levels of detail will result in a noticeable but fairly smooth reduction in detail in the AVCAM files that does not occur in the high-bitrate nanoflash files. This is more apparent viewing single frames or slow motion, at regular speed it can be hard to see unless you're really looking for it. Keep in mind that I am not a video engineer or doing extensive color correction, just someone who has looked at the footage to see if I could see a big enough quality difference with the nanoflash to justify using it.

I think that the AF100 should be at least a little better that the HMC40 in terms of compression or high-motion scenes for two reasons. First, in shallow-depth of field shots, the out-of focus areas will not have much detail and more bitrate can be allocated to the in-focus areas as compared to the HMC40 which is always pretty deep-focus due to the tiny sensor. Second, for slow-motion, which I feel is the area where AVCHD's limitations in codec-stressing scenes are the most apparent, the AF100 will have overcranking such that the max bitrate will effectively be 60mb/s for 1080p60, as opposed to 24mb/s at 720p60 for the HMC40.

Barry Green did a writeup on AVCAM versus XDCAM-EX which has good examples of what the AVCAM codec looks like when stressed, the detail falls off but it does it smoothly:

http://www.dvxuser.com/articles/article.php/25

dcloud

10-22-2010, 10:21 AM

I dont think so... Since the 60p is conformed to 24p the compression will be on every 24 frames

Lpowell

10-22-2010, 11:08 AM

...in comparing AVCcam and AVCIntra shots of a relatively static sean one should not see much difference, the main difference between the codec is one is all I frames the other uses I,P & B frames. When there is little or no movement the lack of all I frames should not show up on AVCcam only when the codec is stressed with quick changes and lots of detail will the difference be come evident.
Yes, and this has definitely been the case with the AVCHD encoder in the GH1 as well. One of the most demanding tests for the encoder is a brightly lit, sharply focused scene with a highly detailed background and a quickly moving object in the foreground. In most other situations, 1080p bitrate will typically drop below 24Mbps. This particular test video uses dim, soft-focus lighting on a featureless background with a largely static subject. It's a test I would expect an AVC encoder to handle without using a great deal of bitrate.

As a test of low-light image quality, however, this footage looks very impressive to me. Panasonic's CMOS image sensors have shown a steady improvement in low-light clarity and color depth, starting with the G1/GF1, and next the GH1/G2. Early test footage of the GH2 looks even more promising and this AF100 clip is the best I've seen yet.

Everts

10-22-2010, 11:15 AM

The avchd implementation on the HMC 150 never broke in PH mode !
Never shot on the lower quality settings though.
And I believe this one wont break either didn't we all just see the Alfa Romeo clip.

Lpowell

10-22-2010, 11:21 AM

The avchd implementation on the HMC 150 never broke in PH mode !
And I believe this one wont break either didn't we all just see the Alfa Romeo clip.
According to the author of the Alfa Romeo clip, he didn't use the built-in AVCCAM encoder, he opted for Nanoflash at 280Mbps.

Yak

10-22-2010, 11:25 AM

The avchd implementation on the HMC 150 never broke in PH mode !
Never shot on the lower quality settings though.
And I believe this one wont break either didn't we all just see the Alfa Romeo clip.
The question is not will AVCcam break the question is will there be a great difference between recording AVCcam or using a I frame codec on an external recorder via HD-SDI. As for the Alfa Romeo clip I believe the they used a NonoFlash external recorder and not ACVcam.

Aneurysm

10-22-2010, 11:32 AM

The stills looks great, but I want to see them try to break the codec.

Everts

10-22-2010, 12:52 PM

Oh sorry guys I missed that .... !

dcloud

10-22-2010, 12:58 PM

you could break it... but the footage would also be useless

maybe footage of pavement on a ground at very high shutter speed could break it. but then again would want that?

Mark Smith

10-23-2010, 05:25 AM

"Broken things" have a way of becoming the next wave of style....

Mark Smith

10-23-2010, 05:37 AM

Back to the codec discussion for a moment..... What would happen if one set out to break both the AVC I codec and the AVC HD codec, what would we find then?? I Don't think the still frame comparisons are particularly informative....

dcloud

10-23-2010, 06:08 AM

I dont think avc i will be broken. Avchd is hard enough tobreak as it is

Duke M.

10-23-2010, 06:35 AM

Its going to be almost impossible to break the high bit rate AVC-I, or Nanoflash I-frame codecs since each picture is a fully recorded individual pixel for pixel snap shot. Color may be sampled at 4.2.2, but once its recorded its there. The limitation then is more for the CMOS sensor glitches than the codec.

Barry_Green

10-23-2010, 08:24 AM

Back to the codec discussion for a moment..... What would happen if one set out to break both the AVC I codec and the AVC HD codec, what would we find then??
Well, you can't break AVC-I. There's no motion prediction or long-GoP nature to it, so every frame is individual, and therefore what happens in one frame has no effect on what happens in the next.

As for AVCHD, I set out to break it and to find its limits. I used an EX1 as the source, as it was the only full-raster 1920x1080 camera I had access to with an SDI output, and I used a separate AVCCAM recorder with SDI input, so it was being fed the highest-res cleanest signal I could get, and then I tortured it. You can see the results in my article in the articles section, in "AVCCAM vs XDCAM EX". I recorded simultaneously with the EX1's onboard codec to see how the two performed.

After that test, I quit worrying about AVCCAM. It's fairly bulletproof.

Yak

10-23-2010, 08:54 AM

Well, you can't break AVC-I. There's no motion prediction or long-GoP nature to it, so every frame is individual, and therefore what happens in one frame has no effect on what happens in the next.

As for AVCHD, I set out to break it and to find its limits. I used an EX1 as the source, as it was the only full-raster 1920x1080 camera I had access to with an SDI output, and I used a separate AVCCAM recorder with SDI input, so it was being fed the highest-res cleanest signal I could get, and then I tortured it. You can see the results in my article in the articles section, in "AVCCAM vs XDCAM EX". I recorded simultaneously with the EX1's onboard codec to see how the two performed.

After that test, I quit worrying about AVCCAM. It's fairly bulletproof.

Barry your test was great and greatly appreciated, but your were comparing AVCCAM to XDCAM EX not AVCIntra. XDCAM EX is long GOP MEPEG 2 not that different than HDV codec. I look forward to a complete test of the AF-100 AVCCAM compared to the various I-Frame external recorders using AVCIntra, PROres and XDCAM. I suspect under most normal shooting situations AVCCAM will hold up extremely well and only start to fall apart under extreme stress. Still it will be important to find out what that point is where AVCCAM begins to weaken.

Everts

10-23-2010, 09:42 AM

For serious production work i'd still consider an external recorder. They said the codec held up in basic colour correction, but i think where the codec will fall apart will be during much heavier correction, not to mention if you wanted to do some serious greenscreen work.

Why and how would in break !? Matthew ...

and how serious is serious work !?

IMO the Native codec, avchd in PH mode, is good enough for capturing any type of work !

dcloud

10-23-2010, 11:32 AM

Barry your test was great and greatly appreciated, but your were comparing AVCCAM to XDCAM EX not AVCIntra. XDCAM EX is long GOP MEPEG 2 not that different than HDV codec. I look forward to a complete test of the AF-100 AVCCAM compared to the various I-Frame external recorders using AVCIntra, PROres and XDCAM. I suspect under most normal shooting situations AVCCAM will hold up extremely well and only start to fall apart under extreme stress. Still it will be important to find out what that point is where AVCCAM begins to weaken.

You knw when you pan the cam andthe rolling shutter starts to ruin a shot? Its waycpast that even more. By the time you worry about avccam compression is the time you wouldnt even need that footage.

Mark Smith

10-23-2010, 11:44 AM

Well I have a 370 in hand and if some one wants to loan me an AVCcam recorder, I'll do a comparison of AVC I 100 and AVC Cam. No Problemo.

Barry_Green

10-23-2010, 03:23 PM

Well I have a 370 in hand and if some one wants to loan me an AVCcam recorder, I'll do a comparison of AVC I 100 and AVC Cam. No Problemo.
Only proper way to do that would be to use the SDI output to send the signal to an AVCCAM recorder (same thing as I did with the EX). You can't set two cameras up and try to shoot on both, because you're introducing a whole range of variables. But sending the SDI out to an external recorder would let you compare the exact same signal as recorded by both codecs.

Barry_Green

10-23-2010, 03:26 PM

but your were comparing AVCCAM to XDCAM EX not AVCIntra.
Yes, but the point was -- I was finding out how robust the codec was. I tortured it. Some of those shots were from high-motion scenes where I was zooming in and rotating at the same time. It's brutal. And AVCCAM held up really well. It'd be hard to think of a scene that would "break" the codec. The EX was used as the camera head, and I simultaneously recorded on its codec to see how it performed, and I was really surprised at how much better AVCCAM held on when XDCAM-EX had puked. But whether I'd been comparing it to uncompressed or Intra or whatever, the basic torture test remains the same, and AVCCAM held up good enough that I am fine with using it. And I'm on record as being the most anti-long-GoP advocate on earth. I hated HDV, and it failed in so many ways in such spectacular fashion... XDCAM-EX is better but still not where I'd want it to be. AVCCAM is pretty solid. I'm not worried about it. For high-budget stuff I'd still prefer an intraframe 4:2:2 codec, sure, but I've been shooting stuff on the HMC40 and AF100 and frankly it looks great. The codec is fine.

Mark Smith

10-23-2010, 03:29 PM

That's exactly what I was thinking: record AVCI 100 to A P2 CARD & pipe the SDI out into an AVC Cam Recorder. Same camera, same signal recording on two different formats simultaneously

Barry_Green

10-23-2010, 03:34 PM

Yep, that's the best way to do it.

Matthew B

10-23-2010, 05:46 PM

Why and how would in break !? Matthew ...

and how serious is serious work !?

IMO the Native codec, avchd in PH mode, is good enough for capturing any type of work !

IMO the 8bit 24mbps AVCHD footage may be a little weaker in the post chain than the 180mbps Prores footage ingested as 10bit. I feel that the latter will give more freedom in post. But at the moment everything is speculation, until I, you, or anybody else has the camera with an external recorder and pushes both the AVCHD and Prores footage in post no one can say whether it will or will not be stronger or weaker. In my previous post i used the word i “think” because i am speculating at the moment as are most people on the AF100 forum.
Now to me ‘serious’ production work is something high paying/quick turn around. In this case, for me, the native fcp Prores files from the external recorder are a much more viable option over the AVCHD the camera records to.
You said that the AVCHD codec is good enough for capturing any type of work, that very well could be true, however will it be capable of handling the colour correction and post workflow of every type of work, who knows.