XDCAM vs CINEFORM quality

Back to post

Comments

LarsHD wrote on 5/28/2009, 2:04 PM
jabloom: "So, in summary, if you are working solely in Vegas, this might be a good approach. But if you need to transfer the clips to someone not working in Vegas, it's not a good idea."


Lars: Well, then just throw in uncompressed AVIs or anything else in the project that is 1920x1080. That's a solution isn't it? If the MXF's are made from for instance 5D2 footage and was converted into MXF, using MXF in the project doesn't make handling the unplanned impossible does it?


Lars
LarsHD wrote on 5/28/2009, 2:36 PM
A few more comparisons to give some perspective..:

1920x1080 29.97 fps


VP8 in Windows XP 32:
==================

PREVIEW / HALF

MXF 35 Mbps.........29.97 fps (full)
MXF 50 Mbps.........29.97 fps
Cineform.................29.97 fps
Uncompr AVI..........14 fps


GOOD / HALF

MXF 35 Mbps.........29.97 fps (full)
MXF 50 Mbps.........21 fps
Cineform.................14 fps
Uncompr AVI..........6 fps


BEST / HALF

MXF 35 Mbps.........17 fps
MXF 50 Mbps.........13 fps
Cineform.................10 fps
Uncompr AVI..........6 fps


VP 9 (with bugs that affect Cineform today which surely will be fixed) in Windows SX 32

PREVIEW / HALF

MXF 35 Mbps.........29.97 fps (full)
MXF 50 Mbps.........29.97 fps
Cineform.................8 fps
Uncompr AVI..........18 fps


GOOD / HALF

MXF 35 Mbps.........29.97 fps (full)
MXF 50 Mbps.........29.97 fps
Cineform.................27 fps
Uncompr AVI..........12 fps


BEST / HALF

MXF 35 Mbps.........29.97 fps (full)
MXF 50 Mbps.........29.97 fps
Cineform.................26 fps
Uncompr AVI..........12 fps


Purerly "mechanically" speaking... MXF *plays* back real well. That makes work a lot more fun.

Lars
LarsHD wrote on 5/28/2009, 2:52 PM
Another nice thing with MXF is that it plays back in the 64 bit versions of Vegas. At *slightly" better frame rates. But with better overall performance and I can use my 8gb of ram.

Cineform in 64 says "offline" in 64 bit...

Personally I think it is time to move over to 64 bit if possible. Now if Cinescore could be made to work in 64 then everything would real fine.

But the fact that MXF runs in 64 bit and Cineform does not, is another reason to go MXF.


Lars
jabloomf1230 wrote on 5/28/2009, 3:02 PM
If the clips are short, uncompressed is fine, once you get something longer at 1920 x1080 29.97 fps, 8 bit, uncompressed file sizes get unwieldy. That's one of the selling points of using an intermediate codec. Intermediate formats like Cineform represent the optimal compromise among the competing factors of file size, portability, preview speed, ease of editing and generational degradation. But it's a minor point. I'm not a Mac guy, but I think that FCP will open a XDCAM HD 422 MXF file, so the Apple side of the street is somewhat covered. Does anybody know this for a fact? Maybe the imminent updates for Premiere Pro and After Effects CS4 will include the ability to natively read and write these files.

Like I said, none of what I mentioned are big drawbacks and if you have Vegas Pro 8 or 9, you get to use XDCAM HD 422 as an intermediate format for free. It's impressive it works so well, given that the codec is just a high bit rate variant of MPEG-2.
cliff_622 wrote on 5/28/2009, 3:19 PM
Wow,...I'm searching the web and having a hard time verifying that this is a true 10 bit codec.

Anybody know for sure?

CT
jabloomf1230 wrote on 5/28/2009, 8:59 PM
You could probably deduce from the fact that people pull the 10 bit HD signal off of SDI on the XDCAM HD cameras and capture that with 10 bit codecs like ProRes 422, DVCPRO HD, Blackmagic and Cineform 422, that the camera's stored format (XDCAM HD 422) is 8 bit. But, it is interesting that although generically, XDCAM is supposedly 8 bit, nowhere is there anything that says what XDCAM HD 422 is. HDCAM SR is Sony's 10 bit codec.
megabit wrote on 5/29/2009, 12:23 AM
You can play back XDCAM, even 422 50 Mbps (or 100 Mbps, for that matter), with FREE XDCAM Viewer:

http://www.sony.co.uk/biz/view/ShowContent.action?site=biz_en_GB&contentId=1167924780439&sectiontype=BC+Tools+XDCAM

AMD TR 2990WX CPU | MSI X399 CARBON AC | 64GB RAM@XMP2933  | 2x RTX 2080Ti GPU | 4x 3TB WD Black RAID0 media drive | 3x 1TB NVMe RAID0 cache drive | SSD SATA system drive | AX1600i PSU | Decklink 12G Extreme | Samsung UHD reference monitor (calibrated)

blink3times wrote on 5/29/2009, 5:06 AM
"Like I said, none of what I mentioned are big drawbacks and if you have Vegas Pro 8 or 9, you get to use XDCAM HD 422 as an intermediate format for free. It's impressive it works so well, given that the codec is just a high bit rate variant of MPEG-2."

I find this all quite interesting. Just a few weeks ago everybody and their grandmother were complaining about how they would never be able to live without Cineform and why hasn't Sony produced their own 'intermediate'.

Maybe this was the direction SCS was trying to point out? I don't think it's a PERFECT fit quite yet, but it does look like a reasonable alternative...... free too.
apit34356 wrote on 5/29/2009, 5:22 AM
Blink, that's about the same line of my thinking on this matter. I don't think SCS is trying to avoid Cineform or other 3th parties. It would a lot easier if SCS would directly offer Sony's free XDCAM viewers on the SCS download site.

I wonder if V9 will handle XDCAM SR as it does XDCAM HD? If so, it offers 10bit.
JohnnyRoy wrote on 5/29/2009, 5:32 AM
> I find this all quite interesting. Just a few weeks ago everybody and their grandmother were complaining about how they would never be able to live without Cineform and why hasn't Sony produced their own 'intermediate'.

It's not that simple and still needs to be fixed. CineForm was introduced as an intermediary in Vegas 5! and shipped along with Vegas 6. For those who purchased Connect HD back in the Vegas 5 days and have been capturing using CineForm ever since, they now have an enormous library of CineForm footage that doesn't work in Vegas Pro 9.0 64-bit. Flat out can't edit it!!! So for new HD shooters this might be a non-issue but for people who have been shooting HD for years, that now have 1000's of hours of CineForm footage that can't be edited, it is a pretty big issue that needs to be fixed.

~jr
farss wrote on 5/29/2009, 5:36 AM
Also as it seems to be broken due to a bug in V9 getting it fixed could fix other issues too. We may also discover that the new adaptive preview feature works extremely well with the Cineform codecs. They seem to have been designed from day one to support this feature. V9 will be the first time we get to see this work in Vegas.

Bob.
cliff_622 wrote on 5/29/2009, 6:37 AM
This is facinating!

Lemme see if I got this right. What we are saying in this thread is that an;

8 bit 422 interframe (long GOP)

Survives multiple rerenders better than a;

10 bit 422 intraframe "lossless" codec? (Cinform)

Wow!,...this is pretty freaky when you look at the numbers."On paper" ,...just going by the specs,...Cineform "should" be a better codec.

Facinating!

CT
farss wrote on 5/29/2009, 7:20 AM
"Cineform "should" be a better codec"

Which Cineform codec?
They make several variants at different prices. The one that ships with Vegas is pretty pedestrian.

Bob.
LarsHD wrote on 5/29/2009, 7:21 AM
It would be great if you can spend a few hours doing the same test that I did.

Try it and see if 6 generations render-render-render-render-render-render also gives you the same result. You need to add some text or some new content in the event so that it actually rerenders.

Compare Cineform 4.8.6 codec with MXF 50 Mbps at 1920x1080. Use color bars and also real footage.

There is a lot of talk... talk... so it would be really intreresting to see if others arrive at similar conclusions after having actually *DONE* the test....

Lars
jabloomf1230 wrote on 5/29/2009, 8:59 AM
Lars,

I have to go back to what Bob (farss) wrote near the top of this thread:

"Unless you've got seriously expensive monitors and test equipment trying to eyeball these kinds of things is impossible. Monitors also introduce banding, noise and artifacts. How they interact with other issues in the image can totally skew what you see."

It is relatively easy to make qualitative statements about artifacts and loss of detail induced by creating multiple generations of a video file, using a specific codec. Unfortunately, until either Eugenia or someone else conducts such a test with the XDCAM HD 422 codec and the results can be compared to identical tests of other common intermediate codecs, all we have are your (and my) positive observations.

J
Laurence wrote on 5/29/2009, 9:11 AM
One more thing Lars:

I'm assuming you did your multiple generations on Vegas 9. It is quite possible that if you did the same multiple generation test in Vegas 8 or earlier, you might find Cineform holding up quite a bit better. Many of us have been using Cineform for quite some time and have never had a problem with degradation over generations. My guess is there is some conflict between Vegas 9 and Cineform that is causing the degradation over generations that you are seeing.
apit34356 wrote on 5/29/2009, 10:24 AM
"Survives multiple rerenders better than a;

10 bit 422 intraframe "lossless" codec? (Cinform)" This not as strange as it seems. If V9 is passing 8 or 32b to the encoder. If 8b, then log math going to 10b can have a small rounding error( a simple mapping of 8->10b by table would be the fastest), the 32 ->8->10 or 32->10b for be math or an index table value thats off. But 10b ->32 --=--32-10-b should be the same( we hope), if one value is off a little either side codec, it would show after multi generations.
blink3times wrote on 5/29/2009, 10:38 AM
"CineForm was introduced as an intermediary in Vegas 5! and shipped along with Vegas 6."

And one wonders if that relationship would still exist today if the cineform price tag wasn't so astronomical. Granted they have come up with Neo scene at $130... and I'm not sure of the prices back in the times of vegas 5.... but I do know that cineform neo hd costs more than the FULL version price of the whole Sony vegas editing platform.... which to me seems to be a bit silly.
cliff_622 wrote on 5/29/2009, 11:40 AM
"...And one wonders if that relationship would still exist today if the cineform price tag wasn't so astronomical..."

Are you saying that Cineform being "broken" in Vegas 9,...was not a mistake but a deliberate act? (because of $$$ reasons?)

I'd prefer to think that it was a simple technology "oversite" by both parties which is on the verge of being fixed soon.

CT : - )

It's SCS that introduced me to Cineform, now I'm hooked on it with allot of my project master archived in Cineform. I'm sticking with 8 untill this is fixed in 9 (Sorry SCS)
LarsHD wrote on 5/29/2009, 12:42 PM
Laurence:
"One more thing Lars: I'm assuming you did your multiple generations on Vegas 9. It is quite possible that if you did the same multiple generation test in Vegas 8 or earlier, you might find Cineform holding up quite a bit better. Many of us have been using Cineform for quite some time and have never had a problem with degradation over generations. My guess is there is some conflict between Vegas 9 and Cineform that is causing the degradation over generations that you are seeing."

Lars:
Thanks for mentioning this Laurence. But I did in fact run this in Vegas Pro 8. Mainly because Pro 9 isn't very stable with 5D2 in render situations and the known issues with Cineform in vegas Pro 9. From a quality point of view I'm not aware that Cineform performs differently in 9 compared to 8. Isn't it only stutter, freezes and "mechanical" things like that, which are the problems in 9 for CIneform?



Additional test today:

I changed the content of my test file to include a wider variation of test charts, bars, stills, moving footage etc.

I now did 20 generations (!) of CIneform and 20 generations of 422 50 Mbps MXF XDCAM....

I could write for hours about all the different details I found here. But one conclusion after 20 generations was:

The MXF file is viewable and isn't that bad. The Cineform has totally fallen apart at 20 generations and looks like a home VHS camera when showing color bars and text superimposed on the color bars..

Text characters etc on the Cineform file are being moved by the codec depending on what color they are or over what color the text i placed. No such things on the MXF file.

Example:
One scene was the camera standing still in a shopping mall. Far away there is a shop with red neon sign saying "ESPRIT". The background is brownish and dark red various patterns.

In Cineform after 20 generations it's very difficult to read this sign. The colors are horisontally smeared an shifted. Cineform seems to have a horisontal chroma problem that MXF doesn't have at all.

MXF at 20 generations can show som light blocks, jpg-artefact looking stuff. This begin to show up on 8th, 9th generation. Typically on almost even surfaces like skies etc.

Comparing 8th generation Cineform is duller, has choma horisontal shifts. Some colors are sideways moved to the left and some not.

Looking at the waveform (luminance mode) shows CF after 20 generations with severely distorted shapes while MXF is till fairly close to the original.

After 9 generations of MXF the overall luminance level is almost identical to the original. Cineform has drastically altered the video levels after 9 generations. And after 20 generations 100% white is now 93% white... Sonys MXF still maintains the correct luminance level even at 20th gen.

MXF shows some light chroma reduction though. Not nearly as severe and disturbing as Cineforms changed, amplified and shifted chroma.

From a pure resolution point of view MXF wins again. And I think this is because of the sideways croma problem that Cineform has. For instance the vertical lines of the stairs of the escalators in this shopping mall, these stairs looks like sine wave resolution test charts... After 20 generations they get blurred and looks out of focus in Cineform. Some of the image data gets shifted sideways, and this clearly affects resolution / perceived resolution.

MXF at 20 shows nice details and resolution per se hasn't really changed from the original..


Again: I think MXF is an amazingly good codec and the various parts of the signal holds together real well.

WIth the quality of MXF they could have offered 100 Mbps and 150 Mbps as render alternatives don't you think?


QUESTION: Can I upload screen shots to this forum? It would be great to actually display my findings here.




Best
Lars


PS. Perhaps Cineform is only meant to be a 1 generation thing, and the altered chroma and luma levels and slight sharpening is something the Cineform developers think is good thing?

If so, I'd say that if I want a file that plays better than original 5D2 AVCHD MOV-file and that plays easier than an uncompressed AVI, I'd prefer just a file that doesn't alter my image data.

Now that I have become a bit more familiar with the alterations that Cineform do, I can see on the 1st render that MXF and the original looks basically the same while Cineform adds shadows etc on text against colors. Neon signs get a little blurred etc.

Examples:

Green text on a red background in Cineform: A black shadow on the left side of the characters is seen. Red text on a green background: A shadow is produce on the right side... And why is the color turned up? Watch the vector scope and see how Cineform is boosting the colors a bit... And why has some kind of sharpening been applied? Or is the sharpening unwanted artefacts? If

Comparing MXF and Cineform 1st generation, MXF definitely messes with the signal a lot less than CIneform.

1st gen Cineform: Black letter against magenta background. Characters gets a litle blurry and appear to move a little left (probably due to the blurriness / transitions between magenta and black (?). This doesn't happen in MXF.

I can't point to any specific phenomenons in 1st gen MXF. But I can with CIneform. And MXF came free with Vegas...

So what is the point in using Cineform at all as a first generation format here?

Add to this that I tried put a 1920x1080 29.97 MXF file and a Cineform file on a low performance external USB backup drive and the MXF plays from within Vegas at 29.97 full frame rate! But the Cineform doesn't.


blink3times wrote on 5/29/2009, 3:24 PM
"Are you saying that Cineform being "broken" in Vegas 9,...was not a mistake but a deliberate act? (because of $$$ reasons?)"

I'm not saying anything. You draw your own conclusions. I'm merely pointing out the fact that this wee little cineform app costs more than an entire NLE. Now to me that doesn't make much sense at all. On the other hand Sony MXF is free. The Edius codec is free. Lagarith (i think) is free. Uncompressed avi is free. The apple codec is free....etc.
jabloomf1230 wrote on 5/29/2009, 4:50 PM
"And one wonders if that relationship would still exist today if the cineform price tag wasn't so astronomical. "

Blink,

David Newman has posted elsewhere that Cineform wanted to continue to license it's VFW codec to SCS (Vegas is permanently stuck at Cineform V2.8), but the version 3 codec needed Vegas to provide SSE2 CPU support and SCS did not make those changes:

http://www.dvinfo.net/conf/1142697-post5.html

Besides, Cineform NEO Scene is not expensive at all and you can find it online from some vendors, as low as $99. I think that you are focusing on Cineform's Prospect HD and Prospect 4K, both of which provide preview acceleration for Adobe Premiere Pro. If you don't use Premiere, you don't need these high end products.

I choose to use Vegas, but I also have to use Adobe products. Cineform is a "life saver" on the Adobe side of the street.

J
blink3times wrote on 5/29/2009, 5:09 PM
Cineform NEO Scene is not expensive at all and you can find it online from some vendors, as low as $99."

Yes.. I noted that above.... but scene didn't come out till a VERY short time ago.... when cineform all of a sudden discovered that there was a market for a cheaper consumer version with the influx of new avchd cams.

And yes... of course David is blaming Sony.... do you really think he's going to hold cineform in ANY way responsible?
farss wrote on 5/29/2009, 6:00 PM
I just ran some tests.
After even one generation the 422 MXF codec is showing an increase in macroblocking, chroma shift and loss of fine detail.
The chroma shift errors might be due to errors in Vegas or my lack of understaning of what's going on exactly. It's also quite possible that the MXF codec is designed to live with these. It's also quite likely that the issues you're seeing with the Cineform codec are due to it expecting Vegas to decode correctly which it's not doing.

The errors are hard to see, I shifted the image by 1 pixel up/across and then back again between generations apart from the initial transcode from XDCAM EX to XDCAM 422. As this was green screen test footage shot as HQ MXF at 25p I applied a CK FX to highlight what is going on when it's switched to Show Mask Only.

The only way I could currently run a valid comparison between MXF 422 in V9 and the Cineform codec would be using our Prospect 4K codec. That of course doesn't work in Vegas so I'd have to do the tests using After Effects. Whatever the results it's hardly going to be fair comparing a $1,500 codec running in a $1,000 dedicated compositing application against something that's thrown in the box in a $500 NLE. Certainly tests by at least one independant post house show that Prospect4K performs better than HDCAM SR and it's the codec of choice for at least one high end post system and several recording systems.

Bob.