Sumix 2/3" 1920x1080 CMOS

[Régine Weinberg]

Well done Jose
ok
I worked with Arri and Aataon for about more than 20 years
of my live, in photo with Hasselblad, Linhof, Cambo too.
Doing shorts and commercials for cinema, working for fashion,
the big ones. Names
All HAS to be digital today to go to print. It HAS to be razor sharp,
Meg Pixel, clean, plastic look It HAS to be as some art directors today even don't remember it has been an non digital age out there.
Time of bad pictures as they like to say. They claim to see the
Meg pix res on screen if it has been 12 MP or 36 ....imagine.

Silly but true, 2001, 2002 I phased out, burned out,
cracked down and had to close my biz.

Silicon started hopefully with a German partner it showed off
Mercedes like , price wise. Hmmmmm

Sumix ok the third party software seems do do cineform quite well
but it is a not easy to take with you product yes and I guess will never
be as there is no real intention to step into and compete with the big names in Cine Cams as some of them are also players in industrial cams.

Red is a bit like Ubuntu.One person drives it the intention to have it out mainly related with Apple, it will be a very handy package the Scarlet and it shows clearly a cool work flow with FCP on Apple.
Nobody else in industry is backing up Indie and low budget film as RED.
Lets face it, Sun is going GPL and community, Red is quite doing the same with Scarlet you get what you pay for.

Hunting DOF can burn much money again, you lose dammed real some light and it will be a huge package again. cumbersome to manage, a pain but with High pro looks.

Why copying the old cine style down to the grain pixelwise?
Why not experiment a new style quick, dirty like Hip Hop and Blues
in full digital film.

Regine

[Régine Weinberg]

please go there
and watch this
http://www.youtube.com/watch?v=dZqxUk_vOIk
http://www.youtube.com/watch?v=wUvxw...eature=related
30 FPG QVGA with the foveon sensor

so what look at the price in US and the $ is low
imaging 4 of them and start with a dremel.
We have to split the incoming image in 4
possible
the DP1 will pump raw out
we have to stitch it
and voila
we can go from 50 to 3200 ISO

attach an 35 MM lens on photo lens we will have DOF and all
eech camera is a full functional system at least

[Jason Rodriguez]

BTW, don't want to hijack the tread, but I thought Jose's post was very interesting, so I'm curious to see how others feel. Right now there is a lot of understandable buzz about higher resolutions, but I'd like to see what the rest of you think, as it seems like everything I hear is always centered around resolution as the key decider for what makes one imaging system better than another.

Do you see overall image quality (this includes dynamic range, sensitivity, color fidelity, DOF characteristics, etc.), functionality (features in the camera, integration with other systems, and overall workflow), or resolution as the most important attribute of a camera system?

Does highest resolution automatically = highest image quality (this again includes dynamic range, sensitivity, color fidelity, DOF characteristics, etc.)?

At what point does the overall functionality of one system out-weight the perceived higher image quality of another system?

Thanks,

Jason

[Jose A. Garcia]

Hi Jason,

I'm sorry if I sounded too biased towards RED or put resolution as the only thing that matters in digital cinematography when I've always thought that dynamic range, color, etc. are very important as well. In fact I think SI cameras give an image much closer to real 35mm film than the RED One, which is known to have its own "cinema look". Besides all that I think you'll agree what RED is promising with the Scarlet is pretty amazing. It won't be a perfect camera. It'll probably have problems in low light situations due to the extra resolution with the same sensor size as the SI2K and it'll probably have the "unique" film look of the RED One, but it'll also start another revolution simply because almost anyone will be able to shoot digital cinema instead of video (that'll be a problem too).

As a filmmaker I love shooting digital because it's fast, easy, cheap and it can give amazing results but I don't really like "pure" digital image. I'm always adding little imperfections in post so it looks more organic and dirty and I hate when digital cameras overexpose lights that could be perfectly natural and soft if shot in film. That's why I started the other thread about film look in digital cinema, because I love that look so much that I don't want it to die. And it's actually dying. It happened with photography and it will happen with cinema. Even people like Spielberg, who says he'll be shooting 35mm film as long as he can buy reel, are slowly entering in the digital world. In fact I saw a digital projection of Indy4 (which was shot in film) and (IMHO) the main reason I didn't like it was because it looked like a TV show. So organic look and dynamic range are two of the most important things for me when shooting or watching a movie and for me SI cameras have that look.

Now, what I don't get is why SI has been offering almost the same, for the same price for more than two years when RED is offering a 3K standalone camera for 1/4 the price of the SI2K mini. I know SI cameras have been really hard to develop (software as well as hardware) but is extra dynamic range and Cineform really worth all that money?

[John Wyatt]

Jason, that's a very interesting (and under-reported) point. Personally, I prefer larger pixels than higher resolution, so that I could film in available light if required. But everyone seems to be chasing big frame sizes even though 2k cinema is likely to be professionally acceptable for some time to come.

[Jose A. Garcia]

I hope this digital revolution doesn't end up being a "who's got the biggest K" war. Let's see if something between 2K and 5K set a standard, being 5K something like "Ultra detailed cinema". If not we can end up with "Super105mm" 25K sensors in five years. Shooting at full apperture in overlighted sets because all those pixels in such a small surface cannot gather enough light. With lenses as big a 4 cameras. Want narrow DOF? How about shooting with a 28mm lens and having 1 inch of the scene in focus and the rest totally blurred?

Ok, that was crazy... But many people said 4K was too much and RED announced 5K for the EPIC.

[Régine Weinberg]

Well
we can say wait and see for 5, 10, 15 years.

Keep in mind cost of reel is going high and higher, and
as with photo the material, eg film is fading out.

In any case you have to scan, to do post and than ??
on reel again maybe that changes in some years.

What we need is a proper work flow
and not fiddling around with bits, peaces, and software
used to control industrial cameras.

SI got it, but did not move , simply put.
Look for photo digital backs, even used
crazy high asking prices. And
PIXEL COUNT war is going on
art directors as young never touched an analog camera
are crying for meg Pixels.
THAT is reality as it happen for ELLE, VOGUE and all

Red is not the holy grain but
a better package as SI as the user base is there.

Years ago there was and still is an Arri,
Aaton and even Eclair community.

Today the Red community is growing.

The Si and Sumix community is not there
so no one is pushing them to move the small steps
we do need to use this camera.

[Paul Curtis]

I've considered these issues quite closely, i've even had a pop at throwing together a camera from sensor heads and hacking around with software so i've got a fairly good idea of what's involved.

Firstly i don't think scarlet is a 'digital cinema' camera, it's a mass produced point and shoot camcorder, albeit with 3k resolution. (I would love to see what post route 'soccer mums' will use :) )

For me, the most important properties of a camera would be

- dynamic range
- DOF
- sensitivity
- colour accuracy
- frame rates
- resolution (so long as it's 2k min)

there is no sensor currently that checks all my boxes. The good sensors aren't available as they belong to sony, panasonic et al. (you cannot get a FIT based CCD that im aware of for example). But sensor technology changes, and it is my hope that the APS-C sized sensors will become more able and this will filter down. But sensors for what we want are a tiny market compared to the rest. I know kodak have a new range of CCDs coming out that improve again so maybe with these. Altasens have their HDR sensors sometime, but i know very little about them, also CMOS need a lot more processing work than CCDs.

There is an argument against 35mm size as well because the DOF is too narrow vs aperture. With an APS-C sized sensor i can use an f1.4 lens with good shallow DOF (same as f2.2ish on 35) and, most importantly, the low light capability of f1.4. If i wanted the same low light capability on 35 at f1.4, the DOF would be razor sharp and pretty impractical. Also in the case of red, the zooms are around f2.8 upwards which is fine for DOF but lack low light capabilities. So i don't really believe 35mm is the best thing.

Digital has the *potential* to outdo film acquisition in time and i personally think that 35mm isn't the holy grail (nor 65). But the APS-C sized sensors in most dSLRs offer a better compromise for day to day practical uses.

i hope this makes sense
paul

[Paul Curtis]

Originally Posted by Jason Rodriguez

BTW, don't want to hijack the tread, but I thought Jose's post was very interesting, so I'm curious to see how others feel.

Jason,

I meant to add to my message above that I think SI should look into the whole APS-C sensor area. I think it's a sweet spot between all the various possibilities and the glass is plentiful. It would certainly differentiate SI from everyone else at the moment. You could take one of the sensors and bin the pixels to increase senstivity and make the dataflow easier off the sensor. I believe i've read in various places of sensors already doing the kind of frame rates needed. They probably all need a mechnical shutter though...

Or Fahard, perhaps summix could do something for DIYers?

cheers
paul

[Biel Bestue]

latitude and sensitivity are the key elements for me, now i'm shooting with hvx with only 4 stops of usable range and it gets me into nerves to shot with that thing, any improvement in latitude is wellcome

anyway, farhad, how would a gpgpu perform in order to encode raw into some kind of lossless codec? would it demand too much power? maybe a mobile computer gpgpu? how does it compete with the actual fpga in the camera?

[Jose A. Garcia]

Originally Posted by Paul Curtis

Or Fahard, perhaps summix could do something for DIYers?

That'd be great too. If Sumix could develop a multipurpose visually lossless RAW hardware encoder with 26pin Micron/Aptina compatibility (to plug ANY headboard) and say DVI out. Add basic internal software to control the sensor and that's all. The board controls width, height and fps. Color correction can be done in post because we're shooting RAW. The hardware encoder encodes just anything that's coming from the sensor so you can have any possible resolution and framerate being only restricted by the encoder speed or the sensor itself. DVI out can be used to plug a LCD with an adapter. Aptina (Micron) sensors are cheap, pretty much standard, fast and have great latitude and sensitivity. If you want 35mm DOF, add an adapter. You can also buy the only Aptina 16x12mm sensor available (with 200fps at full resolution and 2352x1728 pixels) to have real cinema 35mm DOF.

That'd be the most basic DIY filmmaking kit ever. Now we just need Farhad saying it can be done. They already have the lossless hardware encoder inside the 12A2C camera.

Farhad, if you develop that hardware encoder and add something like DVI in, Composite in and a few more, you can actually sell it for multiple different purposes, not just DIY cameras.

[Juan M. M. Fiebelkorn]

what sensor is that?

Never heard of it. Are you somewhat related to micron?

[Jose A. Garcia]

No, I'm not. I know I'm always talking about Micron sensors but I think they're pretty good for their price. I tested their MT9P031 sensor last year for my first digital cinema camera project and found it to be really good in terms of color, latitude, low light...

I'm attaching the flyer for the sensor I'm talking about. The only problem is price. more than 1,000 euro. for the bare sensor and about 10,000 euro. for the demo board. Doesn't say anything about the headboard.

[Paul Curtis]

Originally Posted by Jose A. Garcia

No, I'm not. I know I'm always talking about Micron sensors but I think they're pretty good for their price. I tested their MT9P031 sensor last year for my first digital cinema camera project and found it to be really good in terms of color, latitude, low light...

I'm attaching the flyer for the sensor I'm talking about. The only problem is price. more than 1,000 euro. for the bare sensor and about 10,000 euro. for the demo board. Doesn't say anything about the headboard.

Jose,

This is an example of a sensor that looks good but i don't think is practical for digital cinema. It's really a machine vision sensor and does that very well

- it's 4:3 not ideal but no great problem
- internal 10bit? that seems pretty low processing.
- 7micron pixel size is good, but no mention of fill factor, the sensitivity isn't as good as it could be. Bayer will reduce that even more
- here's the big issue though :16mm x 12mm is in no-mans land as far as lenses go. No S16 lens will cover that, and 35mm lenses will have a crop factor (try and find a 35mm lens wider than 14mm and then find a fast one). I used the 14mm kodak sensor and struggled to find cmount lenses that just about covered it and i even tried a canon 10-22mm efs lens with an adaptor (which worked but impractical for focusing and aperture adjustment)

Hence the APS-C size. It has the lenses needed. No matter how good the sensor, without suitable and quality glass, it's blind. The solution has to be homogenous.

For focus issues something like the birger mount for electronic lenses could be developed so that pulling focus became practical (and wireless focus too). So some hardware to drive the AF and aperture via a separate controller.

It would be a great idea if summix did as you say and developed a processing board for LUT conversion and real time compression. This could even be applied to existing cameras through HD-SDI ports as well if it was designed broadly in the first place. (one box does many things)

cheers
paul

[Jose A. Garcia]

What about this one?

http://download.cypress.com.edgesuit...upa_4000_8.pdf

It's a square sensor but its width is close to APS-C size. It says 15fps at full res. Could be 25-30fps at 1152 or 858 pixels of vertical res.

Anyway we would need a way to control the sensor.