Record uncompressed 14-bit (4:4:4) on your HVX200

Wait a second, there is already a Foveon camera that has video out:

Foveon_digital_camerasystem.jpg


And that Foveon reference design kit has a firewire output.

FOVEON_F7_RDK.JPG


Also announced alongside the F7 and F10 image sensors is the F7-RDK Reference Design Kit, which consists of the higher-resolution F7 image sensor coupled with an IEEE-1394 FireWire port, mechanical shutter, PC Synch connector, 2mm stereo-jack remote shutter release port, Canon EOS lens mount, 3/8-inch tripod socket, and composite video out connectivity. An optional 1.8 inch LCD display is available, the unit is compatible with Type-I and Type-II CompactFlash cards, features a 206MHz Intel StrongARM processor and 128MB of SDRAM, and can output either NTSC or PAL composite video. The RDK serves as a platform for development of digital camera designs based on the X3 image sensors, and allows control of aperture and shutter speed, real-time full-frame video preview, full resolution image capture, VPS image resizing, and the ability to capture 4-second digital video streams. The F7-RDK can output 48-bit S-RGB TIFF files, raw images, dark-frame subtracted images, gray-balanced images and digital video.

A final announcement is the FX3-SDK software development kit. Designed for manufacturers to produce the software their cameras will run, as well as software for the user's host PC to download and process images, the SDK is based on C++ Object libraries. Supported camera hardware includes an Intel StrongARM processor, CompactFlash media, a color LCD display, USB or IEEE-1394 FireWire interfaces, and an I2C bus. Supported host operating systems are Windows 98, 2000, XP and Me, or MacOS 9 or OSX. Image formats supported include Foveon RAW, 24-bit or 48-bit TIFF, JPEG and Embedded Color Profiles. The specific libraries can be found listed on Foveon's site.
Click to expand...


48 bit RGB files !? hmmmmm if someone can mak that work it would be the holy grail.

Hydra Schmydra - if we put our heads together we maybe could make a DVXuser Indie cam that beats everything on the market.
 
Last edited:
Postmaster said:
I guess it is not that easy. I think the mecanical part (grabing the image right after the sensor) could be done by any skilled TV techican, the problem is the software
Click to expand...

If I'm not mistaken its also quite limited in its framerate.
No problem for your Sigma...but if you want to get 60fps its going to be quite a bit harder.
 
Postmaster said:
So why is nobody taking a Foven Sensor camera (like my beloved Sigma) weld some wires at the sensor output and connect the whole shbang to a laptop with andromeda like software?

I guess it is not that easy. I think the mecanical part (grabing the image right after the sensor) could be done by any skilled TV techican, the problem is the software.

Frank
Click to expand...
Foveon chips appear to have been designed primarily for digital still cameras, and because they can't handle high frame rates at high resolutions, they are as yet unsuitable for HD video cameras.

There are three models of Foveon sensors, the X3 4.5 MP (a.k.a. Foveon FO18-50-F19, the X3 10.2 MP (a.k.a. Foveon X3 Pro 10M and Foveon F7X3-C9110), and the the X3 14.1 MP (a.k.a. Foveon X3 Fx17-78-F13). The first two are, I believe, being discontinued.

The 4.5 MP, has a 4:3 aspect ratio and, you guessed it, about 4.5 million effective pixels (1,420 x 1,060 x 3 layers [R, G, B]) over an effective area of 7.1 mm x 5.3 mm (effective diagonal, 8.8 mm). The chip can yield this resolution at 7 fps; at 30 fps, resolution drops to 640 x 480.

The 10.2 MP has a 3:2 aspect ratio and about 10.2 million effective pixels (2,268 x 1,512 x 3 layers) over an effective area of 20.7 mm x 13.8 mm (effective diagonal, 25 mm). This resolution is more than adequate for 2K or 1080p, but it can be sustained at a maximum of only 4.4 fps; at 25 fps, resolution drops to 576 x 384, which, oddly, is a lower frame rate/resolution combination than that of the 4.5 MP.

The 14.1 MP, with about 14.1 million effective pixels (2,652 x 1,768 x 3), has the same aspect ratio and dimensions as the 10.2 MP. This resolution can be sustained at a maximum of only 5 fps; at 30 fps, resolution drops to 640 x 480.

All are CMOS chips. I don't know whether they have global or rolling shutters.
 
an ipod like device to capture video that coinsides with final cut...... apple could take over the world if they made that... i know i would by one
 
IRecordThings said:
Are there any systems like this at all? Why don't we see 4:4:4 in more 3-sensor set-ups?
Click to expand...
Because no video format ever devised records 4:4:4. Video maxxes out at 4:2:2. The Red takes aim at 4:4:4 because the Red isn't a video camera (it doesn't shoot or record "video"), it's a digital cinema camera that shoots digital files.

But aren't all demosaic operations merely interpolating color values? How would this represent true color? Not contesting your point; just curious.
Click to expand...
Well, yeah. The simple truth is that a single-sensor system doesn't sample color at every potential sampling site -- it can't. So all the color information that gets created is created by taking color information from neighboring pixels and guessing (intelligent guessing, supported by neighboring samples) at what the color is "supposed" to be. And, furthermore, it's supported by the very concept of 4:2:2 or 4:2:0 color sampling systems that are used with video anyway -- since the recorded color resolution of video is always less than the luma resolution, some rounding/averaging is inherent.

So demosaic'ing takes the surrounding pixels and samples the various color levels that it's been given and guesses as to what the proper color construction of that particular pixel should be. It's a very well-researched, advanced algorithm and performs very well. But it's because of this process that one can say empirically that it is not sampled at 4:4:4 color -- the color sampling doesn't exist to have that ratio.

But the system works. It's basically the same type of concept as "pixel shift", which works to sample the additional luma sites from spatially offset pixels. Pixel shift needs 3 lower-res chips, Beyer demosaic'ing needs one higher-res chip, but they do the same basic task; a 1920x1080 Beyer sensor would have 960x540 blue and 960x540 red resolution, the same size as used on the pixel-shift HVX, but the single Beyer sensor's individual pixels would be 1/4 the size of the size of pixels on three individual sensors used in a pixel shift arrangement. So, different approaches to accomplish a similar task -- getting more resolution and color information than should be otherwise possible, by working with individual-colored pixels that have not been sampled at a 4:4:4 ratio.

So does the Red actually offer 4:4:4 in sub-4K resolutions without downrezzing, or are they all similarly-affected?
Click to expand...
It's been a while since I dug into the specs so I'm not even sure what Red is advertising at the moment; I thought that Graeme had acknowledged that they're delivering about 3K of resolution -- if not Graeme, then somebody, because I'm pretty sure I read that 3K number on reduser at one point, being the result of both the optical low-pass filter and the demosaic'ing. The 2.7K-to-3K number has also been verified by at least two different resolution tests that I'm aware of, from people who know exactly what they're doing.

So -- at a final image of 3K res (carried within a 4K-sized file) what's the true color sampling ratio? I don't know, but I guess it's possible that at 3K they could have close to 4:4:4.

How long do you suppose it'll be before we start seeing these things in camcorders?
Click to expand...
Lots of hurdles to go before they perfect this technology! Actually I went and looked for some of Graeme's posts on reduser, and he had some interesting comments on Foveon. Basically Foveon has to interpolate on depth because of the limited ability for the longer wavelengths to successfully penetrate the silicon. Most interesting.

Graeme is of the opinion that every sensor design is a compromise, but as for their perspective they believe that an oversampled single chip with a Beyer pattern is the best compromise for digital imaging. A single sensor makes sense for what Red is trying to do; you couldn't use PL-mount lenses in a traditional three-chip system, the lenses are designed to project onto a film plane and a single sensor emulates that film plane idea.

If a Foveon-style sensor were to be developed that offered high-enough refresh rates, I tend to think that would be an ultimate type of option for a single-sensor design because it would be inherent 4:4:4 sampling, but they'd have to do better at getting the red and green through to the bottom layers so they don't end up with a weak signal that needs to be artificially boosted (which introduces noise) etc.
 
Last edited:
Most viewed thread is "I'm afraid I have another 35mm adapter" with like 150,000 views.

Most replied to thread is "Mark and Ted's Excellent Adventure in Drama" from DramaFest.

Most viewed forum........... Cafe Ala DVX.

Later,
Jason
 
If it ever does come out and delivers the goods, I will buy it.

It would be sad if a big company bought the technology with the intentions of shelving it. That would be a big loss for low budget filmmakers.

It's sort of like what they did with the electric car back in the 90's.

Did Juan ever chime in?

Is he still working on the project or is he on a beach somewhere?
 
dont wanna put anyone down, but seriously, if the HYDRA wasnt gonna come out, juan should have announced right away, that way at least if Juan started a new business venture in the future, then people wouldnt be skeptical about his dealings.:thumbup:
 
Wow - people get SO impatient.
When did he announce Hydra? Earlier this year, right?
He said he was aiming for release in the last quarter of the year... so its not like he was exactly dragging his feet.

Doesn't anyone here remember how long it took from the start of the Andromeda project to an actual release?

This isn't anything that should have made people 'skeptical about his dealings'
 
Thought I'd add my own ignorance to the pile. All this mindless gossip is fun!


a) Sony bought Hydra to bury it: Unlikely. Even if Juan released it, I don't see it competing with the EX1. It sounds great, but it does tie you to a computer. And the EX has HDSI anyway.

b) Panasonic bought Hydra to bury it: Hardly. Come on. Indie filmmakers could get great results with Hydra, but you really think it's going to hurt sales of high end cameras? Given its fixed glass, I would have thought the high end market is another kettle of fish.

c) Red bought Hydra to bury it: Hmm. Don't think Red needs to stop something like this. As Barry points out, Red is a cine alternative. It's high end, so again really a separate market.

d) Juan has sold the business due to family, health, lifestyle, or monetary issues: Possibly. Anyone running a small business knows this is an ongoing threat. His silence on that is unusual however, assuming the casual "I've sold the business" email in the reel stream forum isn't horse poo-poo AND the deal hasn't been finalised (note that these two bits of "evidence" seem contradictory: ie "he's in the midst of a deal, so he can't discuss it" but he also casually writes to some guy that he's "sold the business"! Sus.)

e) Juan realised Hydra wouldn't work after all: Unlikely. No reason to necessarily shut shop, and he spoke as if he had already achieved much of the results he was promising.

f) Panasonic bought Hydra to add to the next gen HVX: Possibly. This would be a masterstroke to slow sales on the EX1 and keep the HVX at the forefront of Indie filmmaking, but you think they'd have announced something like that by now, if only to stall people queuing up for the EX.

g) Juan is just busy with life like all of us, and will release Hydra when it's ready: Quite likely.


What it boils down to is that I still think it's coming one way or another, either from Juan, from a new owner, or from Panasonic itself. So, IMO, if there is a sale, it's good for Juan and for us! :D
 
Yea,

Thats a very good business practice to have. NOT

I know several people that have bought HVX's with the intent of getting the Hydra mod to shoot thier movies that are in pre production and are now scrambling to find another camera and are now to far down the list to get a RED in time.

Although I have to say this has been a very entertaining thread!
 
Last edited:
If they were waiting for the Hydra to film, they'll be waiting a hell of a lot longer for their RED number to come up. Not to mention a lot poorer. Ouch.

Funny thing is, if Juan or the new owner happened to make an announcement next week on Hydra availability, it'd probably make a splash twice the size purely because of this monster forum!

Anyhoo, the site's still there [http://www.reel-stream.com/index.php so Juan hasn't taken it all down yet. He just seems to have put in a redirection from .com to the forum.
Oh, crap. Maybe that's my cache. Dunno. I linked to it from Google no probs a moment ago, but clicking on that link directly just now didn't work. Spooky.
 
Last edited by a moderator:
Luis Caffesse said:
Wow - people get SO impatient.
When did he announce Hydra? Earlier this year, right?
He said he was aiming for release in the last quarter of the year... so its not like he was exactly dragging his feet.
Click to expand...
He announced it in March, just before NAB, but didn't say "last quarter", rather
pertierr said:
We will be announcing release dates sometime after NAB, but the target release is during the second half of 2007, allowing for a couple months of beta testing after NAB. Things are moving a LOT quicker this time around, since we now have the experience from Andromeda.
Click to expand...
(post #101) It's not a question of impatience - lots of projects get delayed, cameras included, but in every other case I can think of, it's still possible to communicate with the people involved. They don't shut down their websites, and do tend to give periodic updates - even if they frequently turn out to be optimistic!
 
I just tried your link and got "Error 404 file not found"
Funny thing, the 0 in 404 liiked like the spiral galaxy motif from the Reel-Stream site.
Let's face it folks, Hydra's dead, and Andromeda / Sculptor with it.
At least when CinePorter went belly-up Spec-Com were / are still around doing great things, but this looks to be a total wipe out with nothing to save.
Sad, very sad.
Dave
 
You must log in or register to reply here.
Back
Top