SI + Cineform " Just 2K to anything ? Future ?

[REDHKSC]

The "K" steam is up before NAB2007 !!!,
Check it at : http://www.cineform.com/CineForm-SI_seminars.htm

I prefer to wait to see more from RED + AAA @NAB2007.

Stewart

[IAN SUN]

I am attending the SI workshop in Toronto on Valentiens day

[REDHKSC]

somebody will send the message to CHINA via HKG from Big CHINESE communities for the 2 or 4K to anything.

Remember 2008 world sports event in Beijing and 2010 world expo in Shanghai ?

Stewart
A Gardener in HKG for Greater China

[Paris Remillard]

I just got back from the SI-2K seminar in NYC. They certainly don't have the marketing engine that RED has. I came home from seeing the 4K projected RED footage at NAB NYC and immediatly reserved a camera. But, they seem like good guys putting out a decent product. I was most impressed by the Cineform workflow, though. Namely, being able to edit with unprocessed RAW files, applying LUTs or metadata for white balance, demosaic, and IRIDAS .look files, thus saving huge amounts of render time by leaving the actual demosaic for the edited footage. It seems kinda like the way that Apple's Aperture works. The guy from Cineform(sorry I'm bad with names) mentioned the DALSA demosaic being something like 25 times realtime. I'm not sure where Redcine will end up, but that's a big consideration. Also, I know that what we were looking at was only 2K footage, so the same workflow may not be possible at 4K, however, I do think that I remember another thread where the Cineform folks said that they had 4K RAW working the same way in Premier. It'd be great if Apple could put native RAW workflow into Final Cut the same way they've done with Aperture and even iPhoto. Anyway, I know that most of these topics have been touched on elsewhere on the forum, but I just saw the presentation and thought I'd give my thoughts.

[IAN SUN]

I just got back from the SI-2K seminar in NYC. They certainly don't have the marketing engine that RED has. I came home from seeing the 4K projected RED footage at NAB NYC and immediatly reserved a camera. But, they seem like good guys putting out a decent product. I was most impressed by the Cineform workflow, though. Namely, being able to edit with unprocessed RAW files, applying LUTs or metadata for white balance, demosaic, and IRIDAS .look files, thus saving huge amounts of render time by leaving the actual demosaic for the edited footage. It seems kinda like the way that Apple's Aperture works. The guy from Cineform(sorry I'm bad with names) mentioned the DALSA demosaic being something like 25 times realtime. I'm not sure where Redcine will end up, but that's a big consideration. Also, I know that what we were looking at was only 2K footage, so the same workflow may not be possible at 4K, however, I do think that I remember another thread where the Cineform folks said that they had 4K RAW working the same way in Premier. It'd be great if Apple could put native RAW workflow into Final Cut the same way they've done with Aperture and even iPhoto. Anyway, I know that most of these topics have been touched on elsewhere on the forum, but I just saw the presentation and thought I'd give my thoughts.


Hopefully the this snowstorm doesn't cause a cancellation of the demo in Toronto. The SI guys are scheduled to fly in tonight. I'll post my impressions of the SI-2K footage. I haven't seen any RED footage projected LARGE yet, so I'm salivating every time I think about flying to Vegas.

[Jason Rodriguez]

Glad you guys could make it out to the Seminar :)

Steve Nordhauser, Silicon Imaging's head of marketing and sales is conduction the seminars together with David Taylor who is CEO of CineForm.

And yes, the same workflow you saw at 2K is available at 4K . . . in fact CineForm has been showing dual-stream effects such as real-time dissolves using 4K footage from the Dalsa, and they were using available machinery, i.e., a dual woodcrest workstation (so something like a higher-end MacPro should suffice). So if you're thinking you need to spend $20K for an editing system for real-time 4K performance, you don't. At least if you decide on using CineForm RAW instead of going uncompressed.

CineForm is coming to Mac very soon, so you will have the same capabilities to edit RAW footage at 10-bit resolution inside of Final Cut Pro.

And yes, thinking of the workflow like Aperture for motion imaging (but doing so at the codec level so you're not locked to one application for control of the RAW data) is a very good way to put it. With the QT codec coming very soon, the amount of applications available that you will be capable of working with native RAW in is going to increase quite a bit.

[Jason Rodriguez]

BTW, Just talked with Steve, he made it into Toronto okay, so the show will be happening as scheduled tomorrow.

[Tom Lowe]

Chips are getting so fast and powerful now, 4K online is already a reality, and will be a breeze within a year. I mean, Intel is fooling around with 80-core chips now. 4K may seem daunting to some, but my guess is that soon enough 4K wavelet-compressed post will be the same as 720p HVX/MXF post.

[Jason Rodriguez]

Very true Tom, although not all wavlet codecs are created equal . . . for instance try to decode a single HD stream of JPEG2000 in real-time on even the fastest dual-woodcrest machines out there right now . . . it's not going to happen. CineForm on the other hand, while being loosely based on the same wavelet transform type function as J2K can decode 5+ HD streams simultaneously and do dual-stream 4K on the same box. Then to top it off, CineForm is higher quality than J2K as well.

To place all "wavelet" based compression engines on the same level playing field is like saying all DCT-based codecs are created equal, but as we all know, they're not. They all may start on the same premise, but the methods and outcomes vary wildly. For a given bit-rate, a wavelet-based compression codec will (or at least should) be higher quality, especially visually, than a DCT-based compression codec, making wavelet more efficient at compression in a general sense, but that does not mean that either wavelet codecs themselves will all look the same, or that they will be the same speed in encoding and decoding. So while faster chips will make 4K "wavelet" compression editing in real-time a possibility, it will not do that for the gamut of codecs out there . . . some will be much faster than others.

[Clayton Harper]

I also attended the NYC event at NYU.

The idea of working with preview LUTs on top of the RAW files within the NLE really started to make sense to me when they mentioned that the pilot feature for the SI camera shot 270 hours of footage. I think they said this was like 10TB of data in the Cineform RAW format.

Imagine having to then transcode even half of that after making your selects to a proxy format for editing. Even at DVCPROHD 720p res you'd still need another couple of TB to work with the files. I realize on a big show people are going to be sending their REDRAW offsite to be cloned and transcoded as a dailies process but for something in between film production and ENG type work there is a need to turn footage files a little faster. There is an interesting advantage being able to immediately transfer and edit files in the field on your laptop NLE without transcoding. I have loved doing this with the HVX and would be sad to not have a similar capability like this with the RED. On the otherhand, in FCP you have to do the P2 to QT copy to make things work but this doesn't really take so long. Here's to faster than RT performance in REDCINE.

On a big down note, I learned that the SI sensor chip is still only 2/3" so none of the lowlight and oversampling advantages like the RED.

These things considered, I am still keeping my titanium R close to me at all times.

[David Newman]

I also attended the NYC event at NYU.

The idea of working with preview LUTs on top of the RAW files within the NLE really started to make sense to me when they mentioned that the pilot feature for the SI camera shot 270 hours of footage. I think they said this was like 10TB of data in the Cineform RAW format.

Imagine having to then transcode even half of that after making your selects to a proxy format for editing. Even at DVCPROHD 720p res you'd still need another couple of TB to work with the files. I realize on a big show people are going to be sending their REDRAW offsite to be cloned and transcoded as a dailies process but for something in between film production and ENG type work there is a need to turn footage files a little faster. There is an interesting advantage being able to immediately transfer and edit files in the field on your laptop NLE without transcoding. I have loved doing this with the HVX and would be sad to not have a similar capability like this with the RED. On the otherhand, in FCP you have to do the P2 to QT copy to make things work but this doesn't really take so long. Here's to faster than RT performance in REDCINE.

On a big down note, I learned that the SI sensor chip is still only 2/3" so none of the lowlight and oversampling advantages like the RED.

These things considered, I am still keeping my titanium R close to me at all times.

True it won't have the oversampling, but it will have the low light advantages. The light collection area of each pixel between Red One and SI-2K is the same, so ballpark sensitivity will be similar. Jason can mention the ISO ratings the SI-2K is measuring in at.

[Clayton Harper]

David,

Overall, I was pretty dang impressed by your presentation.

I think the workflow aspects alone might be enough to differentiate these two products.

If I were doing a documentary project and editing and making selects in the field on an extended trip, SI2K would be looking more attractive to me than the RED right now just because of the immediacy and storage implications.

For some reason, the RED frame grabs and footage we have seen so far look less "videoy" to my eye than the sample footage I have seen online from the SPOON feature. These may be technical but they could also be related to artistic or practical choices made by the respective DPs. I imagine we'll have a better standard for comparison after NAB.

[Mike Zinner]

David,
... workflow aspects alone might be enough to differentiate these two products.

Hell no! I want the Cineform workflow for my RED One as well!! Otherwise the RED would be almost useless for me because I am not going to work with DVCPRO offline proxies even if you pay me $1k a day.

REDCODE RAW -> REDCINE -> Cineform RAW

That's the only online workflow I see as of today. Sure, I am happy to wait till NAB and see what the RED team together with Apple can pull out of the hat. But if it does not read "4K RAW realtime workflow within FCP" my Mac Pro will get a copy of Cineform 4K either for PP or FCP, depending on the better integration.

Mike

[Paris Remillard]

Can anyone from Cineform maybe touch on how Cineform will/can work with RED footage? Will one have to convert from RED RAW to Cineform RAW to be able to work with RAW fles in a timeline? Or could anyone from RED touch on if this will be necessary? It's probably too soon to know for sure, but just wondering if you guys are working together at all or thinking about each others products/workflows.

[David Newman]

Unfortunately RED has been quiet on allowing third parties to support their camera (publicly and in direct discussions.) We know we can do it without RED's approval, but it would be better for everyone that there keep their product open. So my hands are tied by RED to discuss how we will be offering support. I can tell you we have asked for direct support within REDCINE allow unprocessed data to be sent directly to CineForm RAW, this would be much faster than the alternatives. We don't need anything custom, just 12 or 16-bit RAW data sent through Quicktime our codec and tools can do the rest. No word back yet, and I don't really expect any until after NAB. So if you want our workflow with tight integration with RED, please politely pester RED of us. :)

[Jason Rodriguez]

Hi Parisrem,

I can let David expound on this a little more, but the REDCODE RAW->CineForm RAW is not a "render" process per-se, but more of a re-wrapping process so it should go fairly quickly . . . I'm not sure how close to real-time you'll get, but it will be *much* faster than going from a RAW format to anything that needs a demosaic step such as all your "normal" YUV or RGB formats.

EDIT: Sorry, David posted while I was still typing . . . didn't realize there was a restriction on REDCODE RAW re-wrapping to other RAW formats.

As far as the SI-2K, the base ISO of the RAW log output at 3200K is 400, and at 5600K is 320 . . . so the camera is definitely sensitive to light. Also at HPA I did a presentation showing 11.3 f-stops of dynamic range from our latest production chips with example images so there's quite a bit the Altasens chips have to offer, and I wouldn't discount them because they are simply "2/3 inch" chips . . . they are a very good, professional, and solid CMOS designed for the broadcast industry, it is not a re-hashed industrial camera chip like you see from other camera manufactuers on the market (not the mysterium of course, but some other CMOS chips that have been hitting the market are not designed for the broadcast industry). Also we do give you full 2K RAW output, so it's more than just another 1920x1080 HD CMOS chip.

[Joe Carney]

What I think is exciting, at least from what footage I've seen and workflow issues considered..the decision as to which camera to use can be based on desired aesthetics, not technology.

Joe C.

[luis bustamante]

David, do you know yet what's gonna be happening in NAB regarding demos and seminars of cineform? I'd really like to attend...

txs

[David Newman]

CineForm has a booth and we will be showing all of this.

[IAN SUN]

Just returned from the SI-2K workshop at PS. I was very impressed with the images produced by the camera. It would have nice though to have seen it projected @ 24P.

I am more impressed by the openness with which questions were answered and the lack of spin.

The Cineform workflow and the fact that they will distribute a free decoder that will allow anyone to view the raw footage (in media player. eek! Currently they are a windows centric company, how long till a Quicktime decoder is available?) made me stand up and say woah! The RED workflow needs to be able to deliver daylies without having to wait a day for demosaic.

Not withstanding, isn't this camera 'half' a RED? @ $12,500 the Mini is very attractive. However, with the SI-2K coming in at $21,000 and projected to deliver at the same time as the RED ONE, if RED delivers with double the resolution, and double the sensor size, why would anyone get the SI-2K?

Make a case please.

[Jarred Land]

Ok kids.. this is a Red-centric forum and its now gone off off topic so im locking it down.. no offense to SI of course, you can carry on this discussion over at dvxuser.