Joakim Sandstrom

Smoke on Mac with 8-core mac pro, 16 GB ram, Quadro Fx 4800, 768 GB SSD (raid 0), Aja Kona 3 + breakout box, Vacom intous 4.

Very little used. Optional Eizo Coloredge CG243W. Subscription still valid. System is currently in Sweden, but can ship it word wide.

Contact

info (at) zermatt.se

if you wish to discuss this further,

Thanks

If you want video that look like film, use a cmos camera! (and 35mm lenses and sensor). Cmos highlights looks much more 'filmic'. CCDs are more futuristic looking ( collateral ) but cmos is more organic or whatever it is. Even the sony ex-1 looks more like "35mm" than the F23 if you ask me. D20 and RED will give you this.

Hey, if you want uncompressed RAW and an insanely fast refresh on the sensor, use the Phantom HD. The only time anyone could detect any screw with the shutter was with lightening bolt strikes.

Mitch - thanks, but I need (hacked) digiprimes. Stereoscopic, dual cameras, lens centers <=70mm apart, infinite focus. Only option at the moment seem to be 2 x hacked f23:s. Unless the HVX3000:s sensor can somehow be hacked into. Must have 60p also.

If it has HD-SDI output then yes you can record that although it may be 4:2:2.

Thanks Phil,

It does have HD-SDI out, single link. Can that do 10-bit and 60p?

Why is it so quiet qbout this camera, HPX3000 with a digiprime must look pretty good,

the specs are almost F23.

About uncompressed data,

I love raid0. All this talk about disk error. I have never heard of a cheetah drive crash.

I have 6 SAS cheetah 15K in raid0, windows x64, I got ~450 MBytes/sec sustained. I read 2K diTTo scanned 10bit dpx files off disk. Material shot on kodak 5201. I've written an opengl program that does color-correction, sharpening etc using GLSL pixel shaders. I am using a Geforce 8800GTX. Debayering in a pixel shader would be trivial. This setup is cheap and robust. Now I want to get rid of film artifacts ( gate movement, grain ) = digital cameras.

But, did not want to hi-jack your thread, this was about the shutter problem.

I think that the only option to Red at this time, in order to avoid a mass "stampede", is to offer a CCD "up-grade" sensor. But it will increase the price considerably, and I don't know if those "loyal" customers will be so loyal after hearing that their "dream" camera is going to be more expensive.

I think the whole look of red is because of the cmos. Cmos chips does looks great. Red is a canon eos-1 for moving images. So I dont think that is an option, you would get a different look ( f23 style ).

The compression in the Red camera being 1/12 looks great!

 

CineForm offers 1/5 and it looks good too.

I hate compression. Never a good thing. mp3 is successful. I hate mp3. Above 24bit 96khz uncompressed is where I hear no difference.

I think that the only option to Red at this time, in order to avoid a mass "stampede", is to offer a CCD "up-grade" sensor. But it will increase the price considerably, and I don't know if those "loyal" customers will be so loyal after hearing that their "dream" camera is going to be more expensive.

Well look at the red forum. People seem overly happy with their images. And the camera produces beautiful results.

I feel there are two camps:

A- People who use the images, in-camera, just like the stuff we see from it today.

B- People who want to capture the most un-compromised quality possible. Think 2001 for example. If I where kubrick I would not want to shoot my elaborate sets, knowing it got 1:12 compression on capture. This fact alone puts the red camera aside from a dual link sdi f23 type of (expensive) workflow

I hate this, I want a 4.5K 2/3 ccd 60p uncompressed now! NAB 2008 ! ? D:

Thanks

Joakim Sandstrom

It seems it will be fine for some people and for some not.

I would worry about

* Stereoscopic shooting, the two sensors being out of sync.

* Motion control work.

* 3D tracking ( boujou ).

I feel red is going for a much broader market ( for now ?) where some issues can be tolerated.

Not competing with for example f23.

Claudio Mirandas red vs f23 pointed out the stuff I was curios about. These small nuanses that no one else

pointed to, rolling shutter was one of them, there where some more,

Q: Shooting into light sources artifacts ( I think you can never have enough backlight )

A: Being adressed

Q: 4K does not look 4K.

A: Seems like 4.5K with OLPF softness is part of what gives that filmic red look. Red recommends adding electronic sharpness.

Q: Compression. 1:12 is way too much. Raw port?

A: They've scrapped the raw port due to lack of interest.

4.5K @ 60p uncompressed bayer, on my disc, this sounded futurustic, so ahead of everything else.

I hate compression

Anyone know if one would be able to directy access the HPX3000 sensor data?

I was hoping the sensor can do 60p, and that the compression is all post-sensor stuff.

It says Picture Elements: Total: 2010 (H) x 1120 (V)

Recording on a PC using a custom, massive SAS RAID0 setup?

It would like that, 2010 x 1120 @ 60 fps >=10bit RGB

If not easily, could someone hack into it and perhaps get this working?

Many thanks

Joakim Sandstrom

Joakim - it's a non-issue technically the size of the lenses. All you need is a beam splitter or halfway mirror in front of the lens and another lens can capture any axis distance at 90 degrees. I could capture stereoscopic imagery with a humans eye pupil distance with large format cameras, if I wanted to.

How would you do that with wide lenses? How would you use for example two UP 12mm in such a config? They produce ~ 90 degrees fov.

http://www.zeiss.de/C12567A8003B8B6F/Embed...2_DigiPrime.pdf

With so much science behind a lens, you don't want mirrors or stuff in front of ( or behind ).

I want light -> zeiss glass -> sensor.

Thanks

Joakim Sandstrom

Dalarö D:

Care to elaborate?

I could go on forever, but generally:

Images are the most powerful form of communication. We trust our eyes and we form our perception of reality based on it. Look at a head, the eyes are two pipelines straight into the brain. A stereoscopic image is so much more powerful than a monoscopic image, it is not even an image anymore, it is a slice of reality. I think the implications for art and science are obvious. For politics, well, take young people for instance, growing up with stereosopic content and the fact that something does not have to be real to feel real... will have tremendous effects on everything. Also negative. Take for example a stereosopic film for kids. They watch the film, being completely absorbed in the presentation. Then some ( evil ) person has managed to hack into the film, suddenly inserting the most terrible images you can think up, people being mutilated, their parents sliced to pieces, whatever awful poop you can dream up. Those kids will to some degree be damaged for life, like people who have experienced war. Stereoscopic content will be that poweful soon.

Thanks

Joakim Sandstrom

3-D when worked properly, it will leave you in awe!...there is nothing that it can compares to it.....sometimes, not even real life.

 

Cesar Rubio.

Cesar,

I cannot agree more. Once this takes off ( by being properly executed ) it will change everything. It will have huge implications on art, science as well as politics. It will completely blur the line between what is real and what is artificial. Combined with the evolution of computer graphics we are about the see the most fundametal revolution in mankind. We will not be able to trust our eyes anymore ;) Apart from the ability to completely hypnotize ( in a good way ) cinema-goers, think about for example remote presence. We will be able to walk around on mars, or walk around on a computer chip, down at the scale of atoms, as if we were really there. Just two out of eighteen thousand examples.

You should read up on current 3D shooting techniques. So much of what you have stated is just flat out wrong.

What? Can you specify? Nothing is wrong. I am 100% right. I don't read I shoot.

Thanks

Joakim Sandstrom

3D research began about the same time as color, but has never had much success beyond the special venue market. Instead of a continuous ramp up as happened with color, 3D has been an off and on thing. It seems to come back every 20 years or so.

The reason stereoscopic films could work this time, in terms of technology, is that it can be done without any flaws. The move from film/artifacts to digital/exactness. Razor sharp, digital capture and projection, zero noise, zero grain, infinite focus, lenses with zero distortion. 60fps or higher playback rate (per eye). Humans can detect ~72 fps. All this is just getting possible today. 20 years ago it was silly to even try all this. The quality of the lenses was enough for it to fail. For example Imax 3D is the exact opposite to all this, it is doing everything wrong.

The only problem that I see with big sensors is that in order to pack more pixels (in 4K, four times more than 2k) and preserve the same PIXEL size for good light sensitivity and dynamic range, they have to make the sensor 4 times bigger!

 

Some manufacturers have packed HD 1080 in 1/3 sensors!

NHK has this 8K camera, using some type of quad 4K cmos setup

Optical format approx. 1.25 inch

7680 x 4320 pixels

60 frames per sec. progressive scanning

http://www.nhk.or.jp/digital/en/technical/pdf/02_3.PDF

Some zeiss glass on that then we are talking.

I guess that must mean they got 4K on a ~ 2/3 inch cmos

I agree with you Joakim, all big sensors are not well suited for 3-D work for the DOF that you mentioned.

Good. But DOF is not the main reason. You could get everything in focus with a master prime 14mm and for example the red one, with enough light. I feel the stereobase is the issue here.

Now you brought up another issue, I found this on your site:

Some professional (more expensive) sensors like the ones used on the SI-2K camera (AltaSens sensor) and the

Red Camera (Mysterium sensor), being CMOS sensors, suffer from this problem also! Although they manage

to "correct" the problem that rolling shutters have by shooting at twice the speed and dropping every other frame

Not matter what they say and do, I prefer a global shutter instead!

That is interesting. Someone comparing red and f23 pointed this out as well. Not many are putting these cameras to the extreme yet. Since stereo should be shot and presented at >=60fps, maybe those cmos artifacts will start to be a real issue. If you are right this would exclude cmos cameras as well.

35mm lenses and cmos sensors = useless. Drastic but perhaps very true if we want optimal stereoscopic cinema.

Thanks

Joakim Sandstrom

Anyone notice how this is another one of those threads where a guy joins the forum and immediately states something incredibly provacative and only semi-coherent, and invariably links it to RED?

I agree this is not so much about red. However, cmos looks better, red is 4500 by something, which is quite a lot of pixels. I would kill for such quality but in stereo, with infinite focus. At the moment two SI-2K mini will produce more accurate stereo. We need a dual 60fps >=4.5K (uncompressed) system with the lenses <= 70mm apart. Not much will be able to compete with that footage.

Thanks

Joakim Sandstrom

Live action that's going to be shown on a large screen needs to be shot with a smaller stereo base.

The Russian Stereo70 system uses a 26mm stereo base.

I believe you want wide lenses (only) for optimal stereo. Premium quality wide lenses that does not curve lines. Digi prime is the only lens I can think of that has the quality and is small enough for stereo base <=70mm. It is 95mm but can be hacked. But what (wide) lens would you use with even smaller stereo bases. Well this is exactly my point here - stop making larger sensors, we need more like a (dual) 1/3" 8K sensor and a zeiss 80-90 degree lens pair for it.

Thanks

Joakim Sandstrom

I'm not worried. It will never happen.

The more powerful a technology, the more skilled operator you need to use it. In the same sense, the stronger the medium, the more it hi-jacks you brain. That's why you'll totally hate it until it is done right. But when it is, which may take a while, nothing will be able to compare. Perhaps flying in space. It will completely take over vision and hearing. The screen will disappear. People will be having trouble separating these screenings from reality. Add to this the fact that a stereoscopic image gives a much stronger imprint in peoples memories, it is no longer an image you are watching.

you have to use a different layout to that used by small cameras like the SI or the Pace modified F950s for James Cameron.

This is exactly my point. The Pace Fusion 3D camera is the only one doing this right. And I am sure they will upgrade it to use the F23.

But everyone is crazy about 35 mm sized sensors and 35mm dof in general. Which is so 1930. H*tler used 35mm cameras and lenses.

This is 2008. Think >=4K stereoscopic cinema with 60fps playback rate. And you need to shoot these stereoscopic films with real wide,

premium glass lenses. All these other systems you mention produce sub-optimal stereo because they fail to do all these things right.

Having the lenses and sensors ~70mm apart ( Pace 3D camera ) is the way to go for all systems. You would even want to have a smaller stereo base to shoot scale models which has real-time non-repeatable action ( otherwise you can do this 2-pass ).

Master prime 14 was just an example but here it is

http://www.arri.de/news/newsletter/article.../mp_14_150.html

Thanks

Joakim Sandstrom

RED has been used on 3D films, so there's no reason why can't be used in the future. Agreed smaller cameras are more handy for 3D work and the SI mini seems to be fitting in there.

Well IMAX 3D has been done for quite some time as well...

Human eyes are approx 70mm apart. Two hacked red one:s + for example two master primes 14mm + hacksawing the lenses, you

will still not get 70mm stereo base. You will get some doll house effect. Future sensors must and will be smaller than 35mm.

Thanks

Joakim Sandstrom

The future of cinema will be stereoscopic.

Steroscopic cinema requires a stereo base of 70mm. Or you will get wrong scale ( doll house effect )

It is hard to get two 35mm sized sensors and lenses that close.

Also, you want everything in focus, ie the exact opposite of what seems to be an obsession with "narrow 35mm-like dof".

This means all 35mm cameras, 35mm sensors and 35mm lenses will be useless once the stereoscopic revolution takes off.

I think Red Two should be a dual 2/3" cmos or foveon, but digiprime-compatible.

Thanks

Joakim Sandstrom