What file format are you guys capturing to???

I am see alot of drop out when I capture... these weird splotches...

but the original video is perfect


thanks!

hdr-fx1

A little more information would be helpful, Brent. If you are using Cineform, you can only capture to m2t, and then convert to .avi. You can instruct Cineform to convert the m2t to avi on the fly, but it's still doing a conversion.
How fast is your system? Hard drives? CPU? RAM?
It all comes into play.

Brent,

If you are trying to edit m2t files natively, you will experience dropped frames and sporatic playback in Premiere Pro 1.5. I used MainConcept's MPEG Pro 1.0.5 originally which offers no HD acceleration.

I just downloaded the Premiere Pro 1.5.1 update (which incorporates CineForm's AspectHD as the HDV codec of choice) which allows realtime playback but at a price... CineForm's claim of a "visually lossless" codec is totally bogus (gosh, I'm so 80s).

Upon further investigation, I found out that Adobe locks AspectHD to "medium" compression, as to where, CineForm's AspectHD 3.0 allows you to specify different levels of compression for their CFHD codec. I tried the highest quality setting and the footage was still too heavily compressed for the visual effects work I was trying to do in After Effects.

The "realtime" conversion from m2t to CFHD was lagging by about 2 minutes when capturing 60 minutes of footage. I use a 3 GHz Pentium 4 w/HT enabled, 1.5 GB PC1066 RDRAM and my capture drive is a 1.6TB RAID.

I ultimately ended up downcoverting the m2t files to dv, editing the dv footage, creating an EDL and then importing the m2t files to replace the dv files in the EDL. This actually allowed me to work faster as the computer was not being taxed to death by handling HDV footage until the very end. I exported the effects footage as an uncompressed avi file for After Effects (FYI, 2 hours of uncompressed 1080i is approximately 800 GB).

I think to truely edit HDV you'll need a dual processor system (I haven't tried a dual G5 w/Final Cut yet). Dual Xeon's are becoming less expensive and with dual-core chips available this summer, editing HDV will probably become as easy as editing DV is today.

You don't need a duallie, but it helps. The DV workflow is a good one though, and one that I use regularly.
Unless you are working with extremes types of footage, the Medium setting is best in Cineform. I'm surprised that with Aspect though, you weren't getting great playback. That's a surprise. I'm getting 29.97 on a laptop using Aspect.

Xander,

sounds like you had some technical problems as visible artifacts should not be occuring (note: some users had conversion problems in earlier versions, which has since been addressed with free updates.) The CineForm codec has been used for the most demanding of film work; it is the first compressed HD codec to used entirely as a digital intermediate, no other compressed codec can claim this (see the cover story of January's Post magazine.) Maybe what you need is a 10bit solution, 8bit workflows can introduce more artifacts than compression itself. The "Dust to Glory" project used compressed 10bit throughout (including effects work.) Visit http://www.cineform.com/technology/HDQualityAnalysis10bit/HDQualityAnalysis10bit.htm to see why CFHD can claim visually lossless.

On a technical note: Adobe ships no Aspect HD (acceleration) components, only the codec ships with Premeire Pro 1.5.1. Aspect HD is an optional quality and acceration upgrade to the base components shipped with PPro1.5.1.

David Newman
CTO, CineForm

<<<-- Originally posted by Douglas Spotted Eagle : I'm surprised that with Aspect though, you weren't getting great playback. That's a surprise. I'm getting 29.97 on a laptop using Aspect. -->>>

Thanks Spot. The new Aspect HD 3.0 offers even more acceleration. On my under-spec'd 2.8Ghz Prescott P4 (Dual DDR helps), I can comfortably play back 3 streams of Sony HDV footage (up to 5 streams in short bursts.) All with only 1Gig on RAM and a single 60GB (old) 7200RPM IDE drive.

David Newman
CTO, CineForm

Mine works great with only a 2.4 HT pentium 4, 1 gig ddr 400, surprised how well it runs. I capture the m2t files first then convert to CFHD avi. As a matter of fact it is starting to feel a bit like DV editing :)


Mike

Sorry, I should clarify my experience with AspectHD:

Douglas -

I had no playback problems with the converted m2t files using AspectHD (or Adobe's use of the codec). The footage played back fine on both my workstation and my laptop.

David -

I can't afford an HD 10-bit capture/playback card, so I have no choice but to use the native 8-bit colorspace that HDV records in (I only use firewire :) ).

I read the article you linked to and that's what made me decide to download the demo version of AspectHD a few nights ago. This short film I shot last week is my first foray into HD(V), so I'm learning.

My beef with AspectHD is compositing in After Effects. The m2t files (after I uncompress them) key super clean with minimal fuss with the Keylight plug-in for After Effects. The CFHD compressed files using both the "large" and "medium" quality required a significant amount of work to minimize the artifacts surrounding the actors on greenscreen... a productivity loss in my book. And the CFHD didn't look anywhere near as good as the uncompressed m2t keyed footage.

I think the "visually lossless" claim may be valid on recompression of source footage for documentries or narrative videomaking (i.e. "Dust to Glory"). But for effects work, I personally wouldn't use CFHD-based files.

I don't know of any technical problems I could have encountered being that the installation of the software went without incident and the codec shows up in all my video apps. I even read the manual and used HDLink for capture and conversion.

By no means am I trashing your product, but the claim of "visually lossless" misled me to believe that the CFHD codec was better than native m2t. In my experience, it's not. CFHD just allows me to edit HDV in real-time which is really all that CineForm promises. I just took it one step further to fit my needs and it didn't work.

Sorry for being vague before guys... My bad...

I have a dellpoweredge 400sc .. p4 2.8mhz
2gig ram
3 terrabytes of externals
editing Vegas5


I think I fixed my first and immediate problem...

Using Cineform to capture raw m2t files... it (by default) was set to medium quality--- I put it on HIGH and it seemed to have solved that problem...

The problem now is that the files are so clunky and slow--- it makes it so painfully hard to deal with for editing...

I tried to go into Cineform and "convert to avi"
and here is the Error message I get:

http://www.resmedia.com/error.jpg


Please advise and thanks to everyone for the help!

Xander, you experience doesn't much that of other customers doing green screen work. But I do acknowledge that chroma keying is the demanding of excerises, we even added the Chroma Noise Filter for HD10U users to key JVC's camera (which has chroma issues.) However, once your keying is done there is no other format more suitable for your post work. On last note, when hear of difficult footage we would love a sample of the orginal m2t file to see whether the compression can be enhanced further. The CineForm codec is based on a mathematically lossless transform (i.e. zero loss) but the data is then mindly quantized (to bring the data rate down), so the quantization can be tailored to any source. So I would be very greatful for 5 seconds of M2T data on CD or FTP.

Brent,

It sounds like you are trying to edit the original M2T in Vegas (yes that is very slow.) Add that HDLink will not convert because error message seems to indicate a installation issue. Visit the support link on www.cineform.com and search on "conversion trouble", the information there should fix this for you.

David,

I can't give you any "motion" footage until the film is released for distribution (May '05); however, I will upload a frame or two of the keyed m2t footage vs. CFHD-converted footage in the near-future to my site and you and your engineers can take a gander at what I'm talking about.

I did talk with a video colleauge of mine who has more HD experience than I. He said I must have "eagle eyes" to notice any sort of compression with AspectHD, but he did agree with the keying scenes that there were artifacts, so it just can't be me.

Is AspectHD using wavelet compression with data compaction? Similar vein to Matrox's DigiSuite? If so, it's the wavelet compression that's anti-aliasing the edges making it harder to get a clean key around the edges of the actors.

How is this all going? I am interested to hear what is happening.

David -"The CineForm codec is based on a mathematically lossless transform (i.e. zero loss) but the data is then mindly quantized (to bring the data rate down), so the quantization can be tailored to any source"

Is this where the difference between HIGH quality and STANDARD quality Cineform capture/convert come into play? Or in other words, what really is the difference between the two capture settings?

Ken,

The difference is very subtle. There is less compression applied to the "large" mode versus the medium settings. The compression difference between the two modes applies mainly to the high frequency image components (this how all image compressors work.) There is less compression of the highest frequences in large mode.

As I understand it the High Frequency is the part of the image that deals with edges which give the detail. Is this true? If so then that might be why the blue-screen problems.
Does this capture mode have any effect on CC ability? Would it help the image not fall apart so fast?

Yes detail is carried by high frequency information. That is always compressed in video system based on the human visual model. We can see low frequency quantization easier than high frequency quantization. This is how all human modelled compression works from DV to MP3. A visually lossless compression can reduce the bit-rate by compressing the elements you cannot see. Uncompressed HD is around 100MBytes/s, CineForm compression is around 12-14MBytes/s -- it looks the same, but of course mathematically it isn't. There are plenty of books on this subject -- just type "video compression" into the search field on Amazon.com.

Blue-screening is a little more difficult for any compression system as it does not follow human visual models -- it is not the human eye that is doing the key, it is some third party mathematically model. So this very much depends on the logic of the key. Use a good keyer on a compressed source and it will work great (rememeber most sources are compressed in some form.) We re-did some tests after this thread and found chroma-keying of our sources works fine on a good keyer. We have many customers using CineForm compression of compositing work. Unfortunately I haven't yet received sample data from this particular customer to see if anything can be done to improve the situation (I hope that happens.) As you quoted, we have a mathematically lossless transform, so we can adjust to store any sequence.

> Does this capture mode have any effect on CC ability?

CC = Color Correction? Yes the codec is very good at color correction and preserving the quality over generations. Using "Large" capture mode versus medium has not much impact on color correction.

> Would it help the image not fall apart so fast?

The whole point of CineForm compression is to greatly reduce multi-generational losses. You can see this analysed here : http://www.cineform.com/technology/quality.htm