(Ed: Referring to Cineform codec; this was moved as the beginning of off-topic discussion in Cineform thread) It has the best price-speed-quality ratio, yeah. Avid DNxHD's codec is serious too, but Avid is not.

c'mon eugenia. I use avid, i love avid.for all the nle's i've tried, only avid works time after time, once learned has the most efficient workflow, and has allowed me the fastest editing i've ever done. avid has gotten a bad rap. give 'em a break.

You didn't get what I wrote. I didn't write anything about Avid's editors, I talked about Avid's DNxHD. Avid is NOT serious about the PC implementation of their codec. The UI for the codec comes with a hideous visual bug that it's so glaring, that makes me think that either not enough testing goes into this thing, or no one cares over there about it.

I even talked to an Avid employee about it months ago, he said he will pass it on, but nothing was fixed!

So yes, I stand by my words to say that while Avid is a serious company, they are not serious about their PC DNxHD codec.The PC port feels to be a side project that some engineers put together during lunch time. And given that we are talking about intermediate codecs over here, that to my eyes, tells me that it can't compete with Cineform. Not because they don't have the manpower or money or expertise, but because they don't believe in their codec. The Cineform guys do.

100% with you Clyde.

so if you want a REAL implementation of DNxHD codec, you have to :
- use own avid software ($2000) + add $9000 to have SDI output.
- use broadcast implementation of DNxHD (telestream pipeline HD for $11 000), or EVS Xs video server ($80 000)
- use third party implementation (merging Vcube with avid option + AJA card for $11 000)

so $11 000 minimum to play out realtime SDI... thanks Avid.

if only they make the free Decoder multithread able, we could use premiere + blackmagic for less than $1500! this is their politic of codec widespreading lol.

Eugenia...
I gotta admit, you make a valid point. If you're referring to the chronic dropdown box overlay in the properties window. it's pretty amazing that that persists. For a while, I thought the problem was with my video card, but, now I know better. Isn't it a QT problem, tho'? It persists into other apps.

c'mon eugenia. I use avid, i love avid.for all the nle's i've tried, only avid works time after time, once learned has the most efficient workflow, and has allowed me the fastest editing i've ever done. avid has gotten a bad rap. give 'em a break.
I consider Avid a top company, right up there with General Motors, IBM PC/Laptop Division, AIG, and the California State Government.

Started this new thread as a home for some posts that were a hijack of a Cineform thread. All posts prior to this one were pulled from the Cineform thread.

What are we talking about here? Can someone please explain?

I've never heard of any of this before, which certainly doesn't mean it's not valid. But I run DNxHD 220x (ten-bit) on a Pentium 4 w/ 2 gig and no external hardware. It works exceedingly well. I haven't experienced any codec problems.

As for the quality of DNxHD, I've read tests that show it does a somewhat better job than Apple's ProRes. I've seen 220 compared to uncompressed and could not tell the difference.

That said, I believe Cineform is a better codec than either DNxHD (or ProRes). But that the 220x version is a legit mastering quality codec for 4:2:2 material.

But please do share what you guys have discovered, and if there is a problem with the PC implementation of DNxHD thank you very much trying to point this out to Avid.

Peter...

IMHO this is a rather trivial thread. Not sure why it was broken out...really not worth the trouble. Some people love to slam Avid, I love to slam vegas. In the end, none of it really matters, but, the faithful always react rather violently.

I think what eugenia is referring to is the OPTIONS window when DNxHD is selected for export, on a PC. The codec options window has shown a corrupted display for as long as I've used MC. It's no big deal except that the problem has persisted over several versions.

There's always a lot of intensity related to any discussion of NLE choice. Avid has never let me down, unlike other NLE's I've used. But then, that's my personal experience and others feel differently. The one thing I can't accept is the release of a new version of software that brings my production to its knees. Premiere has done that and Vegas has done that. Guess I've learned to test the new release of any mission critical software.

On the subject of Cineform and Avid, Avid will allow importing of a Cineform DI, but, it will transcode it, as it will with anything except DNxHD, into the DNxHD DI. The beauty of the Cineform DI is that it imports and needs no other transcoding, like importing XDCAM does. The Cineform DI has new features that allow "raw" image processing via the metadata. I've found this to be a great workflow that doesn't damage the original files. I'd love it if Avid adopted the Cineform DI as one option for an internal (to Avid) DI.

Okay, I found the bug. After some settings are changed, "Avid DNxHD Codec" gets kind of chopped off and reads "Avid DN." Never noticed it before. Also surprised that it hasn't been fixed. But it has absolutely nothing to do with the performance of DNxHD.

Cineform DI sounds nice. I do think that Avid is going to have to come up with some type of native 4:4:4 and RAW solution for MC and Symphony. DS is $60K.

But for some people in this thread to make it sound like you need $11,000 to effectively use DNxHD is unadulterated BS. I pinch myself everyday at Avid's amazing performance on my old PC. Every now and then I try using Vegas for more than the ocassional capture, and I'm startled by how slow it is by comparion.

You guys crack me up. On our Avid here at Jacobs/NASA I have over 10 projects with about 70 interopened bins between them linked to a total of 72 hours of HD/SD footage right now and not even one instance broken media offlined clip. Not too serious.

I have lost zero hours of editing due to downtime.

I'm in the process of training 2 Vegas guys on Avid. One of them realizes that he will be able to get a job at a post house a lot easier if he knows Avid.

We're having fun cranking out a lot of good serious work.

After all, it is about the work.

You need to read the original thread to understand what was talked about. This thread should not have been split up like this, because the original thread contained a link to my benchmark testing which actually showed the bug mentioned here and compared Cineform to Avid. Anyway, that's the article: Eugenia’s Rants and Thoughts Blog Archive Intermediate Codecs: the face-off (http://eugenia.gnomefiles.org/2008/09/15/intermediate-codecs-the-face-off/)

And if anyone of you works for AVID, please fix the visual bug. I mean, look, I am a software gal myself (was working as a developer before I became a tech journalist), but this is such a glaring 100% reproducible bug, that I really don't know how it got slip through the cracks and got shipped in that state. To me, that screamed "not tested".

...the original thread contained a link to my benchmark testing which actually showed the bug mentioned hereEugenia and Bill, that's EXACTLY why it WAS split off -- the other thread was ONLY about Cineform performance and it was hijacked with off-topic discussion of Avid. Comparisons are fine when they help to illustrate a point, but this discussion was far afield from that. This Avid forum is your place to discuss Avid and its virtues and woes if you wish to do so. If you find it trivial, then simply ignore the thread.

Now back to your Avid discussion...

Eugenia,

I very much appreciate that you took the time to do this comparison, but I want to reply to some of the statement you made and how they apply to Avid.

"Normal users would usually use it in 1/4 HD preview windows on their home PCs, but professionals are more likely to use full HD preview second-monitors, so I thought it would be good to test both."

Are you aware that Avid can playback FULL resolution (only 8-bit color though) on any properly confgured modern computer? I do this all the time on a six year old Pentium 4 system w/ 2 Gig of RAM, no RAID, no hardware accelerater. And MANY effects and color corrections playback full time w/ no quality hit; no frames drop. Sure I can change to 1/2 or 1/4 resolution in Media Composer, but I have no reason to. No other NLE that I've come across can do that.


"I should also mention that Avid installed its codecs to work under the Quicktime .MOV container rather than the AVI or MXF containers, which has a speed impact under the PC (just because Quicktime under Windows sucks goats)."

Like I mentioned above, Avid is blazing fast. Using a Quicktime wrapper does not seem impact performance in Media Composer.

As for your interface bug, I can replicate it in Vegas. But it seems like a Vegas problem. Attached is a screen grab from doing a DNxHD export from inside Media Composer. You can see that "Avid DNxHD" gets truncated but that's it. I've also attached a screen grab from a TMPGEnc Xpress 4.0 export dialogbox. There are no problems with it at all.

Now as for the most important part of your observation, I have to do more testing. I just don't have an answer as to if DNxHD does not properly handle interlaced material. I have to think that's not the case, but you do clearly show your results. If they are indeed a bug or just setting that needed to be swithced, IDK, b/c I use DNxHD with mostly progressive footage.

Thank you for your input and I hope I cleared up misconceptions about Avid' and I will look into what you've discovered and hopefully get an answer.

HTH.

Hi peter...


Thought I'd jump in and add that the pictures you posted must show the OPTIONS menu from the latest version 3.5 release. I say this because the display problems in earlier versions were much more corrupted than this shows. So, it would appear that Avid is already working on this issue. Thanx for your help.

Nope it's 3.1.3. Maybe it has to do with the verison of Quicktime that's being used. I'm using 7.5.5.

Eugenia,

Do you know what version of the DNxHD codec you were using? The latest version is 2.0.

Here is a link to the package (the download is under "File Attachments" near the very bottom of the page):

DNxHD (http://avidtechnology.custhelp.com/cgi-bin/avidtechnology.cfg/php/enduser/std_adp.php?p_faqid=77958)

I have used both 1.9 and 2.0 with the same UI bug appearing.

>Are you aware that Avid can playback FULL resolution

Generally speaking, my statement was correct. Most editors aren't doing that, and I was testing under Vegas anyway.

>Like I mentioned above, Avid is blazing fast. Using a Quicktime wrapper does not seem impact performance in Media Composer.

Not on Vegas. Vegas is very unoptimized with ANY MOV file (regardless of codec inside), it requires AVI to be working in its full potential. So Avid's DNxHD is very slow under Vegas. Besides, the "native" container for Windows editors is AVI, not MOV. Avid *should* provide both! Vegas is not the only PC editor that has trouble with MOV btw.

>I can replicate it in Vegas. But it seems like a Vegas problem.

I don't believe this. And the reason I don't, is because:
1. NO OTHER codec UI ever had this bug, but only yours. And I have installed a number of third party codecs.
2. All Vegas does is to invoke the DLL for the UI, it doesn't mess with it.

So, sure, there might be some conditions that exist that create the bug when invoked from within Vegas, but I believe that ultimately that UI bug is Avid's.

>I have used both 1.9 and 2.0 with the same UI bug appearing.

When you performed the multi-generation test, was it with the 2.0 codec?

>Generally speaking, my statement was correct. Most editors aren't doing that, and I was testing under Vegas anyway.

I just wanted to let you know that full resolution playback is very nice to have and certainly a reality inside Avid (w/o external hadware).

>Vegas is very unoptimized with ANY MOV file (regardless of codec inside), it requires AVI to be working in its full potential. So Avid's DNxHD is very slow under Vegas. Besides, the "native" container for Windows editors is AVI, not MOV. Avid *should* provide both! Vegas is not the only PC editor that has trouble with MOV btw.

I think the reason why Avid chose MOV is because Avid runs on both Mac and PC. Since DNxHD in MOV does run well in the PC version of Media Composer, I'm not sure that Avid will take the time to make AVI versions. But I do see your point.

>I don't believe this. And the reason I don't, is because:
1. NO OTHER codec UI ever had this bug, but only yours. And I have installed a number of third party codecs.
2. All Vegas does is to invoke the DLL for the UI, it doesn't mess with it.

But if you look at the window that surrounds the dialogbox for the DNxHD codec options in Vegas, it's the same size as the ones used for all the other Avid codecs. I believe the window that the codec options are displayed in is determined by the application, i.e. Vegas in this case. It looks like Vegas does not realize that DNxHD needs a different size options window than all the other Avid codecs.

And how can you explain that the window displays properly in TMPGEnc? The only certain condition that's different is that it's a different program. And you can see that TMPGEnc uses a slightly different Options interface window than Avid uses, which indicates to me that the Optoins interface window is also determined by the calling application, not solely the codec.

Eugenia, again I appreciate your help and please feel free to respond.

Let's be clear about what problem we're really complaining about here. The graphic problem is a trivial one. The claim that DNxHD isn't suited for a PC is more serious. It is my opinion, as you may have yours, eugenia, that the problem lies within vegas. I have used vegas since version 3. vegas has NEVER handled Quicktime well, while other NLE's, such as Edius, do fine with QT. Likewise MC handles QT on my PC just fine. vegas claims to be codec agnostic, but, in reality it isn't. TmpGenc, Squeeze, Procoder, and MPEG_Streamclip, they ALL handle DNxHD without a hiccup, not even a burp. Vegas' problem may well lie with the fact that Vegas still uses the archaic vfw, rather than the contemporary DirectX.

for you to judge DNxHD on the basis of your experience, which is limited to vegas, is not only unrealistic, but, somewhat iconoclastic, eugenia.

It would seem your anger at Avid is related to the fact that DNxHD is wrapped in QT and not AVI. Again, eugenia, I think you're being rather parochial.

>When you performed the multi-generation test, was it with the 2.0 codec?

1.8, I think, at the time. The article is not new you see.

>for you to judge DNxHD on the basis of your experience, which is limited to vegas, is not only unrealistic, but, somewhat iconoclastic, eugenia.

No, it's not. I am a Vegas user. I don't look into moving into another editor. More over, 90% of my readers are Vegas users as well. That's what we/they care about. If Edius is able to play DNxHD without problems, good for it. But I ain't gonna use Edius even if someone puts a gun onto my head: Edius problems (http://eugenia.gnomefiles.org/2008/04/19/canopus-you-suck-or-software-sucks-part-9/)

Look, I am a tech journalist, but the article was published on my personal blog, not on my publication (OSNews.com). The way it works in the business, is if someone wants a review of their app or their hardware, is to send me over a free copy. Both Sony A Look at Sony Vegas Pro 9 (http://www.osnews.com/story/21606/A_Look_at_Sony_Vegas_Pro_9) and Adobe A Look at Adobe Photoshop, Premiere & After Effects CS4 (http://www.osnews.com/story/20552/A_Look_at_Adobe_Photoshop_Premiere_After_Effects_CS4) have done so (read the reviews and you will see that I write it as I see it). I can't possibly get my hands on the rest of the popular editors without spending thousands of dollars from my own pocket (especially for Avid's editors/HW).

>It would seem your anger at Avid is related to the fact that DNxHD is wrapped in QT and not AVI.

No. I am unhappy with Avid (not angry, I don't care, I use Cineform) because of both AVI, the UI bug, and the slowness on my editor. EVEN if all the blame is to go to Sony, I still can't use the damn thing. And that's the real bottomline. I have none of these problems with Cineform, because by using AVI: 1. it doesn't have a UI bug, and it's fast to decode. So even if the blame is Sony's, it's Avid who decided to not go with AVI and not support Vegas (and other PC editors) properly.

As for the bug being Vegas', I am not convinced. I will take it up with Sony and see what they say.

>1.8, I think, at the time. The article is not new you see.

It was actually 1.9.

Eugenia,

If you could be so kind as to do your generational test with the latest version of the codec, that would be very helpful.

Thanks.

Unfortunately I don't have Cineform installed on this machine anymore (neither the exact frame used in that tutorial). Therefore, the only test I can do is test DNxHD with another frame, and then observe the generation loss with itself, rather than with other codecs. If another codec is required for comparison, I can test it with Lagarith in YUY12 mode (not in RGB mode, as that would be completely lossless).

EDIT: email sent to Sony.

... Therefore, the only test I can do is test DNxHD with another frame, and then observe the generation loss with itself...

That's what I'd want to see. The comparisons across codecs I'm not conecerned about.


P.S. If you'd be willing to also run the a generational test of 175x on progressive 23.976 that would be great. It would help determine if it's a problem with movement between interlaced fields only, or if movement between progressive frames also causes the problem.

Ok, sometime this weekend then.

Eugenia...

My deepest apologies for leaving you with the impression that I meant to down-play you. That is not my intention. I've read your article and found it informative and well written.

I would, however, suggest that you could best serve your readers by being as technically accurate as possible, and leave personal opinion and emotion out of the story. One of my biggest complaints with this entire NLE business is that there is very little unbiased reporting going on. And that's true for every NLE vendor, not just the vegas faithful. In the end, the unwitting noobie gets pulled into emotional arguements with very little foundation in fact. I guess the old axiom always applies....caveat emptor.

World peace ;).

I can assure you, there is no Vegas bias. Read my review of Vegas 9 on OSNews and you will see the hammering they get too. My benchmark article was as technically correct as possible too. It's just that it used Vegas as its platform, because that's what I use at home. As I explained above, I can't possibly use all the world's video editors to test a codec. It's impossible. You just pick an editor, and you go with it. Vegas is as good as any, because, even if it has shortcomings (e.g. MOV support), it has other benefits that other editors don't (e.g. Cineform or Lagarith AVI wouldn't work as well on FCP -- if at all). So it all balances out at the end. The various articles have to be written within logical constraints. Even if you had written my benchmarking article, and you have decided to use Edius or Avid, the FCP/Vegas users would have found a way to tell you "it's unfair" about other aspects of the article and the codecs used and how well or not these codecs are supported under the editors of your choice. As you can see, it's impossible to have an 100% objective result if a codec is good or not in all aspects, because all codecs NEED a video editor or encoding UI (e.g. even the command line ffmpeg), and they have to work through the constraints of these tools, and at the end you have to have a conclusion within the constraints of these tools. I hope I am making some sense. :)

I agree with you, completely. I am a huge proponent of the Cineform DI, preferring it over DNxHD. As I said earlier, I would hope, however unrealistically, Avid incorporates the Cineform DI as part of their DI package, without a need to transcode to DNxHD on import.

Test is done! Download the PNG files and the results in text form, here: http://eugenia.gnomefiles.org/images/avid-dnxhd-generational-test.zip
I can't possibly upload the original files though, they are huge. 2.0 version of DNxHD was used. It seems that 2.0 behaves much better than the 1.9 version in terms of interlacing content. Only interpolation visibly shows a generational loss (and no, this is not because you lose half the res with interpolation, as this happens for all generations, including the original file).

But for some very weird reason, the progressive results are worse than when in my first article (might be scene-dependent). On my original test with DNxHD 1.9, it wouldn't have more than 20,000 pixels of difference, but this time it did over 35,000.

Eugenia,

Thank you so much for re-doing the test. I looked through the pictures and read the Readme. I also compared the performance to your original test.

As for the progressive footage's worse performance than in the previous test, it could be like you mentioned b/c the image is different. The new one looks very challenging. But I also have read that DNxHD 175 had some banding issues that prevented "ER" from using some Red footage transcoded to it for broadcast. Avid said they have worked on the issue, but who knows for sure. That said, the marks for the progressive multi-generations still look very good, assuming I understood the tests properly.

Thanks again very much!

Thx Peter. BTW, regarding the visual UI bug, apparently it does not happen on older versions of the codec (e.g. v1.2.0), which leads me to believe that the bug is Avid's and not Vegas', that happen to get reproduced under specific conditions.

Eugenia,

I'm looking into the UI bug some more. It's the least I can do ;). Hopefully I'll get an answer and post back soon.

Eugenia,

The best I've been able to find out is that it's an issue with Quicktime. Apparently there are different ways for a program to interact with Quicktime. In the case of TMPGEnc and Media Composer, they do it in a way that doesn't cause the error. In the case of Vegas and Sorenson Squeeze, they use a different manner which causes the error.

As you probably know by now, clicking on the middle of the bottom border will cause the drop down box to appear, but that's not a real fix.

I'll let you know if I find out anything else.

Eugenia,

I just want to let you know that Avid released a new version of it's DNxHD codecs.

Here's a link:

Knowledge Base (http://avid.custkb.com/avid/app/selfservice/search.jsp?DocId=263545&Hilite=)

Realize that if you use the 2.1 version, you will be decode previous versions of DNxHD, but the files you encode will not run properly on computers w/ the old codec versions installed.

I'm not sure if this version fixes the drop down box problem that you mentioned. But it does alter how interlaced footage is decoded. Which was problem YOU had noticed.

HTH.