RED CAMERA Prelim. Specs

abeljerrod said:
Zem...
Guess I forgot to include the "Sarcasm" icon :laugh: .... it was a joke regarding the RED logo looking like the "EASY Button" that all of the Staples Television ads are inundating us with.....
Click to expand...

ok, i guess the problem is that i dont watch tv then (that and that i live on thee other side of the pond... and that my english sucks :)
i should have realized its a joke anyway though, seeing that you're from abel and that you probably know your stuff ;)


Luis said:
I think I'm going to change my signature to:
RED CAMERA: You don't HAVE TO shoot 4K, but you can if you want to.
Click to expand...

Luis,
the point is that a good 4K sensor is more expensive than a good 2K sensor, and that even regardless of money the best 4K sensor with 60p has less sensibility, more noise and other problems than the best 2K sensor.. hence my point of rather focusing on a a 2K chip with great performance than to jump for insane resolution that nobody can take advantage of. also, the internal processing power and bandwith is going to be much higher at 4K.. can you imagine the amount of processing needed to resample 4K raw to a great looking 1080p signal in real time? (i can't, but i know from others that debayering quality depends a lot on the method used). on the other hand, outputting 2K is pretty easy and the debayering could be conveniently done on a cheap pc offsite (of course i'd still want a live 1080p feed, but only for preview and editing).

anyway, i could be all wrong about this, but building a 4K camera and targeting for low cost filmmaking just seems a bit stupid to me.
zem
 
Don't forget that a single CMOS sensor camera using Bayer must use a sensor with a multiple of 2k resolution to achieve 2K resolution output. The ARRI D-20 sensor is 3k, for example, to achieve uncompressed 4:4:4 1080p. How the image is processed is very important. The Dalsa camera uses a 4k sensor, but the method for processing the image is quite cumbersome (it essentially has to be "developed" after capture). We don't yet have enough info to judge the merits, or otherwise, of the RED sensor and processing technique. But I do generally agree that many users would be happy with less spatial and temporal resolution in favor of greater sensitivity and dynamic range. I'm reserving judgement on RED's approach until the facts are in. That happens when we can see a video stream on a big screen. Everything else is speculation, but it's exciting to have another player in the market.

Greg
 
zem said:
the point is that a good 4K sensor is more expensive than a good 2K sensor, and that even regardless of money the best 4K sensor with 60p has less sensibility, more noise and other problems than the best 2K sensor.. hence my point of rather focusing on a a 2K chip with great performance than to jump for insane resolution that nobody can take advantage of.
Click to expand...

I think there's nothing insane in 4k. It is already in DCI specs and film scanners and printers with that resolution are coming. Not to mention the digital projectors for theathers. When RED camera will be selling (2007?), 4k will be normal thing for big budget productions.
By 2007 you will also be able to edit 4k material real time with normal work station.
Canon's still camera has already double the amount of pixels and still pictures are rarely shown in a size of a movie theather screen.
(Once again that camera is also feeding 70Mpix/sec so we just need to record that data...)

And it might also be that it will be cheaper to manufacture 100 000 identical sensors than design and manufacture 90 000 small ones and 10 000 bigger ones.
If you want to shoot 2k, you just use the center part of the sensor (easily done already with cmos cameras) and the pixel size is the same than with smaller 2k chip.
With having smaller area of the chip used, you can also use smaller and cheaper lenses.
 
Graeme_Nattress said:
And 4k bayer is about doable on todays computers too. 4k RGB is a bit tricky with the data rates though.
Click to expand...

errrm, 4K in 12bit RAW bayer at 60p is more than 1 GB/sec (read gigabyte per second - or did i mess up something here? 4096x3112x12/8x60 /1024/1024 =1094MB/sec). at 16bit RGB that would be 4.3GB/sec.
24p is still 440MB/sec in 12bit RAW.
how excactly this is doable with todays computers and how it should be affordable for the indie filmmaker in 2007 escapes me.
zem
 
toke lahti said:
By 2007 you will also be able to edit 4k material real time with normal work station.
Click to expand...
i strongly doubt that (see above)..

Canon's still camera has already double the amount of pixels and still pictures are rarely shown in a size of a movie theather screen.
Click to expand...
.. and even cheap 35mm still cameras have had more than twice the image resolution as 35mm movie format. moving images need much less resolution than stills, if you try to make a 60x45cm print from a PAL originated image it looks crap.. but i've seen (and made) PAL film-outs on 10x6m screens and they looked surprisingly good.

(Once again that camera is also feeding 70Mpix/sec so we just need to record that data...)
Click to expand...
?

If you want to shoot 2k, you just use the center part of the sensor (easily done already with cmos cameras) and the pixel size is the same than with smaller 2k chip.
With having smaller area of the chip used, you can also use smaller and cheaper lenses.
Click to expand...
yeah, but you loose depth of field, light sensivity and probably other things in comparison to a full size 2K sensor.

personally i think that the D20 is prettty perfectly designed for high end digital film making, with 3K RAW... make a camera with the same features for about 1/5th of the price or one with slightly lower specs at 1/10th and you could have a winner.
and while you're at it, make a version with 1080p RAW on super16 sized chip for less than 10'000.. that's enough pixels for me ;)
zem
 
I agree, the D20 is very nice. The lenses make it even that much more buttery.
Here are some frame grabs of a test run of the D20 - from a recent
article by Stephanie Argy in American Cinematographer Magazine.
Curtis Clark, ASC used very few lights for the scenes.
In fact, the exterior scenes of the man walking down the street were
all shot using available light from street lights and street signs.
The salon scene used only two Kino Flo ParaBeams and one Pocket Par.
Clark was noted as saying that the camera was pulling details from shadows
that his light meter had indicated he wasn't.
ARRID20FRAMEGRABS.jpg
 
Sure it's doable. Didn't say it would be cheap, but 440MB/s is not impossible with a nice big fibre array. 8bit RAW would be withing the realms of what my edit suite could handle, or compressed RAW would certainly be, whether it be 8bit or 12bit. The jump up to 4k RGB is a large one though, and offers no quality advantages over RAW....

Graeme
 
Graeme_Nattress said:
Sure it's doable. Didn't say it would be cheap, but 440MB/s is not impossible with a nice big fibre array. 8bit RAW would be withing the realms of what my edit suite could handle, or compressed RAW would certainly be, whether it be 8bit or 12bit. The jump up to 4k RGB is a large one though, and offers no quality advantages over RAW....
Click to expand...

yeah, 440MB/s is doable, but not affordable for 99% of us, and not easy to handle on-location shots (power supply and noise issues)... 8bit RAW would be a complete waste in the first place (who would shoot 4K with 8bit precision?).
my understanding is that you'd want to edit with offline proxy files anyway, as editing with RAW files would be very processor taxing (at least thats my experience from RAW stills) and you'd still need a lot of fast diskspace, while DVCPROHD proxies could be edited on a cheap workstation and the original footage stored at a save place (says somebody who killed a full Raid0 the other day out of stupidity).

anyway, i guess the opinion differs on how this camera should be designed depending on how people wanna work with it. personally i'm a do it yourself for cheap kind of person, so i'd like a camera that shoots full raster 1080p or 2K and that i can record uncompressed to a convenient affordable medium, like a mag with four 2.5" drives.
others might prefer a 4K cam with dual HD-SDI out for recording to a HDCAM SR deck that hooks into their existing workflow, but costs 10 times more.

zem
 
zem said:
i strongly doubt that (see above)..
Click to expand...
Well, lets assume that this cheap indie production accepts 4k24fps for delivery format. So we need to handle that 440MB/s. By 2007 we can take 8 pcs of 1TB sata disks with $4k, make them raid0 and have 5 hours of disk space with enough bandwidth for realtime preview.
If camera would use 2:1 losless compression 4 disks would be enough.
Or we could implement mild visually losless 4:1 compression and with 8 disks we could get 20 hours of footage online, so there you can cut almost a full feature.


zem said:
.. and even cheap 35mm still cameras have had more than twice the image resolution as 35mm movie format. moving images need much less resolution than stills, if you try to make a 60x45cm print from a PAL originated image it looks crap.. but i've seen (and made) PAL film-outs on 10x6m screens and they looked surprisingly good.
Click to expand...
So all we need is fewer pixels in a sensor and more pictures out of sensor per second.
Dataflow from sensor is already fast enough.

zem said:
?
Click to expand...
Canon 1Dmk2N is pushing data out of sensor with speed of 70 megapixels per second. It can do so only little less than 3 seconds at a time, this limit comes from recording to CF, so once again dataflow from sensor is already fast enough.

zem said:
yeah, but you loose depth of field, light sensivity and probably other things in comparison to a full size 2K sensor.
Click to expand...
But if the option is 2/3" HD/2k sensor then the pixel size would be about same.
And you'd have an option to shoot 4k when possible.

zem said:
personally i think that the D20 is prettty perfectly designed for high end digital film making, with 3K RAW... make a camera with the same features for about 1/5th of the price or one with slightly lower specs at 1/10th and you could have a winner.
and while you're at it, make a version with 1080p RAW on super16 sized chip for less than 10'000.. that's enough pixels for me ;)
zem
Click to expand...
D20 with 1/5th of the price! That would be a perfect guideline!
And maybe with 1/10th of the price volumes could raise so much that it would be still profitable...
 
Jarred Land said:
-4k, 4520 x 2540 Mysterium CMOS sensor, true S35mm sized.

-60p frame rate at full resolution.
Click to expand...

Well...here's the problem though, or maybe it's not a problem:undecided , however the camera only shoots 60P at full rez and not 24. This might change as development continues.
 
Policar said:
Oh, and my magic price fairies tell me they're shooting for between $25,000-$75,000.
Click to expand...

What is this now...were playing "The Price is Right"? :cheesy:

Well...The price is WRONG ...BOB!:laugh: (I had to say that sorry....to excited)

My guess with the amount of R&D cost and overhead I think it might end up being more like $25,000–$75,000 X 2. (Safe Guess Huh?:) )

Good point Jarred!
 
Shiloh Arts said:
Well...here's the problem though, or maybe it's not a problem:undecided , however the camera only shoots 60P at full rez and not 24. This might change as development continues.
Click to expand...

Of course 60 fps is the MAXIMUM frame rate of the sensor, not the only frame rate.
 
I just found this article: http://www.dpreview.com/news/0512/05121201new_chips.asp

Previous attempts to do this on-pixel conversion have required far too many transistors, leaving too little area to collect light. The new designs use as few as three transistors per pixel, reserving nearly half of the pixel area for light collection. First tests on the chip show that at video rates of 30 frames per second it uses just 0.88 nanowatts per pixel--50 times less than the industry's previous best. It also trounces conventional chips in dynamic range, which is the difference between the dimmest and brightest light it can record. Existing CMOS sensors can record light 1,000 times brighter than their dimmest detectable light, a dynamic range of 1:1,000, while the Rochester technology already demonstrates a dynamic range of 1:100,000
Click to expand...

Much better dynamic range, more light and less power consumption. Maybe Jim can pursade them to make some 35mm chips :)
 
On second notice, at the bottom of the page, maybe this has already been addressed, but it says their full frame sensor is 24.5 x 13.5. I thought full frame sensors had to be 36 x 24. As a matter of fact, their sensor is actually closer to an APS size sensor, the type of sensor most commonly found in current DSLR models.
 
a 35mm motion picture frame is smaller than a 35mm stills film frame. The dimensions stated are correct (near enough) for motion picture 35mm film.

Graeme
 
You must log in or register to reply here.
Back
Top