Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

High Speed Red 1000 Fps

Jannard said:
I'll have to correct you... you have made assumptions that have never been given out or spoken. Guessing doesn't make it fact.

Jim
Click to expand...

Hi Jim,

sorry if I was in a rush and put in to many assumptions. I justed wanted to demonstrate poeple how - lets call it "hypotethical" - sensor parameters could extrapolate into high speed on 5K camera with a 4x3y format chip (which is what many hope for on EPIC).

I agree that it is wrong if people read my message and take it as a "fact" (which was not my intention, sorry if it could be read like this). But the math applied is technically correct and applies basically to the RED ONE too, if you put in its variables. At least its what it seems to me. A new sensor design may change this too, but what we know from other sensors and the RED ONE Mysterium sensor, the math is basically leading the right direction.

One parameter which I did not talk about was the speed of the RED CODE encoding engine, which is currently the bottle neck in RED ONE (thus we did not get 60fps 4K RED CODE). But it has nothing to do with the sensor physics, right?

Again, my intention is not to derivate any numbers from given (nor changing) or speculative specs, but to show the relation between a high resolution and high speed sensor on the one side, and my personal wish for a high speed RED camera on the other side.

I've come into this "alternative imaging" subject by shooting highspeed on my own. I was camera operator on many major commercial shootings using high speed cameras in the past. These cameras always worked by shooting RAW, so thats what paved my way into REDs thinking - long before RED was existing in terms of a camera system or even idea to make one. So as an early adopter for RED ONE I'd also like to see RED complement the market with high speed cameras.

High speed comes as a side effect when resolution goes up - or when you make a dedicated special sensor for even more high speed. I was shooting 1000fps HD in the past, so there is *sometimes* a requirement for this. We even did "making of" of shootings with 1000fps. We did even shoot high speed with moving cameras, e.g. on a panther etc. Its different filming and its a niche thing, but an amazing one. And I am tired of renting those stuff, I want my own one...

Thats a reason why to me e.g. a PAL sized "region of interest" window high speed readout of RED ONE Mysterium sensor is a valid wish - and why I asked in the past if there is chance to see a "custom mode" to achieve exactly this kind of functionality from a RED ONE.

Its not my only use for RED ONE, but an occasional one.

While most existing high speed cameras come from a technically / industrial background (observing & controlling machines and robotics, dynamic analysis, motion capture etc.) most of them lack image quality and the adequate processing, nor something like RED CODE to crack down the amount of data to be "moved" between two shots.

In the past we shot about 10-20 takes a day with film. Its annoying work to get your film camera ready for a next 1000fps shot sometimes...

With most other dedicated high speed cameras we shot roughly 75-100 shots a day, depending on the shot and its preperation time. But essentially the download time from camera into computer still turned out to be the limiting factor. As high speed shooting is often a try & error game, you often need many shots to make it "perfect" - or you simply make more shots in your given time frame to get more perfect results. And now you can have your customer at your side to watch and see the results - immediately.

Jim, if one can make, you can make it!

Think positive :)

Cheers,
Axel
 
Axel,

Of course you could always use a Phantom HD with a CineMag. If you ask RED to continue to add capabilities to the EPIC concept then at a certain point the cost of doing so will begin to rival the cost of getting a Phantom in the first place. And that camera is available right now.

Just sayin'... ;-)
 
Hi Mitch,

I think you are right that putting in more high speed increases cost. The question is how much.

Phantom is not a "mass product" compared to the RED concepts. So if RED is getting "some" more high speed capabilities out of their basic designs, it might end with a good balance between possibilities and price.

Currently we release most shots still in SD, so having a RED ONE shooting an SD window would already give me some decent high speed capabilities. Its not necessarily 1000fps, but for food shots etc. you rarely exceed the 250-400 fps range anyway. Thats nearly within reach on a RED ONE, it should be within reach on an EPIC.

I know the guys who made the Phantom sensor. They also made the ARRI sensor. They made also an 8k * 8k 10,000 fps sensor. They get paid for the design, thats it. The chips - in the end - will differ only slightly in price. For wafer production you pay (once you have your masks done) per area. Clearly high speeds result in a lower yield rate, so classifiying the sensors will result in less amounts of "high speed capable sensors". Take the 1280x1024y Micron / Aptima sensor (formerly developed by company that Micron bought). Its a sensor designed for 500fps. You can find cameras doing 1000fps with it. They select them to get this speed. Micron doesn't recommend anything beyond 800fps for the beast, though... This thing runs on 66 MHz system frequency. That isn't really fast nowadays, where Intel CPU cores exceed 4 GHz in speed.

A typical wafer will cost you roughly 800 US$ with a minimum setup of like 12 wafers per job. Sensors are comparably "massive" structures compared to todays CPU cores with their 45nm structures. I've seen 12 MPixel sensors smaller than 1/4.5". Thats a fairly small structure compared to a 35mm sized sensor. Clearly its more noise, but not as much as one could derive from the size difference.

Phantom and all those other dedicated high speed cameras are a pure niche product, and therefor the high prices they have. When RED enables higher speeds in their product line, it'll be a consideration up to certain fps speeds to choice a "standart" RED camera rather than a dedicated high speed camera. Remember, many typical 35mm cameras have been able to shoot like 120 to 150 fps. Why not have a RED do this or more (at a smaller window) when a user wants it?

Axel
 
The moment you begin to select a smaller window for increased speed you begin to lose visual acuity. This is not simply a relationship of the resolution of photosites, but also based on the OLPF that is put in place in design of the frequency response for the whole chip. RED is fairly aggressive in their low pass filtering. This eliminates aliasing effectively and helps give the "creamy" look that Jim speaks of, but it also means that under lower windowed resolution the image will drop in resolution accordingly. I would question just how usable an image one could get out of an NTSC or PAL windowed extraction. I would think it would result in a fairly fuzzy image, even for standard def.
 
Mitch,

the RED team always spoke of an effective resolution like 3.2K from a 4K shot - after deBayer. So that was the aim behind the 5K, to have 80% effective = 4K resulting resolution. At least thats what I have been told.

Given than, we should shoot about 1/4th bigger than the desired target and should be "as fine" as with 4K.

Anyhow, I wished for a custom mode. As you operate cameras like the Phantom, you know the flexibility of those systems. I would love to see this on my RED ONE. I was told it might take too much FPGA space to implement, though it might not be possible to implement it alongside the current functions. Jim once announced a "video firmware" at some point. Why not consider a "custom mode firmware" for such special purpose needs.

I do not see RED concur with the Phantom in the ultra-high speed sector, that is clearly impossible. But for some basic special effects this might be interesting to customers. Clearly we would need to change the firmware and make a black shading. Isn't that something we get used to?

What do you think?

Axel
 
I think Red One is still an amazing camera, and sure for many people like myself the main camera for everything, what I hope for the future is that RED include this fantastic machine in the path of sensor upgrade, I am sure many people will jump in ecstasy around the world.
 
Just shot this video with the Phantom Flex 4K and it was super addicting. I'm really hoping that red release a 1000fps camera before i pull the trigger and buy a Phantom :-). Need the dynamic range of the Dragon!

 
Last edited:
You must log in or register to reply here.
Back
Top