Missing Scarlet Info

[Gerald Menclik]

incredible job Justin - working out all the Nov 30. informations for us!

 

[Joseph Hutson]

Felix, in our ScarletUser podcast with Ted that we recorded a few hours ago, I asked Ted a number of questions. I chose about 3 or 4 questions, specifically from this thread you started.

The podcast will be made available in the next 2-3 days...

 

[Steven Caesare]

Hi Steven, you are actually confusing sensor read times with frame rates. Epic's sensor read time is indeed 5µs (microsecs) and I'm sure actual shutter speeds will be able to go quite high.

Justin, yeah, you and I have posted on this matter in other threads as as well...

I know Jim wrote "microseconds" there, but I suspect that may be a typo. Milliseconds are "thousandths" of a second, wheras microseconds are millionths of a second.

Remember Jim is comparing to a film camera, with a spinning mirror/shutter. If 4 microseconds were really the case, then the spining shutter in the film cmn would have to rotating at 250,000 RPM. I'm pretty sure that's not the case. The ~250 RPM that 4 milliseconds would require is much more likely.

Not to mention that I seriously doubt that we could see the skew represented by 4 millionths of a second difference between a film camera (5 units) and a Red One (9 units) as compared to the 1/48th of a second (~21 milliseconds) of the frame exposure if we are talking microseconds. However, the difference of 5ms vs 9ms as compated to 21ms of exposure is likely much more visible...

I could certainly be wrong though.

-sc

 

[Justin O'Neill]

I know Jim wrote "microseconds" there, but I suspect that may be a typo. Milliseconds are "thousandths" of a second, wheras microseconds are millionths of a second.

Remember Jim is comparing to a film camera, with a spinning mirror/shutter. If 4 microseconds were really the case, then the spining shutter in the film cmn would have to rotating at 250,000 RPM. I'm pretty sure that's not the case. The ~250 RPM that 4 milliseconds would require is much more likely.

I think you still may be confusing sensor read time with shutter speed.

On a film camera the 4 microseconds refers to the time it takes for the leading edge of the shutter to move from the top of the frame to the bottom. If the film is 21.95mm × 16.00mm, 4 microsecs refers to the time it takes the leading edge of the shutter to travel 16mm.

 

[Justin O'Neill]

Steven, I saw the post where you say Jim referred to milliseconds. I'll try to find that. Mitch Gross also referred to skew in microseconds, however:

The spinning mirror for reflex viewing is also a mechanical physical shutter to the gate. When the shutter is covering the sensor, the chip can then activate the electronic shutter, and then it really doesn't matter how long the read/write time is as long as it can happen while the sensor is blocked (a very long time). Think of the read/write time to b equivalent to the time it takes a mechanical shutter's leading edge to pass from one edge of the sensor to the other. Then the whole shutter is blocking the sensor for a very long time and finally the trailing edge of the shutter reveals the sensor from one edge to another for another short time window, after which the sensor is exposed to light for a long period. On a traditional 180° shutter at 24fps, the time the sensor is fully exposed and fully covered is just shy of 1/48 of a second, while the read/write time that the edge of the shutter moves across the frame is measured in microseconds. The Phantom HD GOLD does the read/write in just a microsecond, but it is also designed as an extreme highspeed camera and it costs a pretty penn more than the RED One. The RED One takes much longer, but I cannot recall the time at the moment.

Here is another post from Mitch referring to microseconds:

All the pixels get switched on in a rolling fashion. Start on the top line and they start switching on one after another, in microseconds. The second row begins the process shortly after the first -- it does not wait until it is complete. The same with the third, fourth, etc., in a cascading action. This read cycle can take a matter of milliseconds if not microseconds. It depends on many aspects of the camera design, and not really with the sensor design itself. It is about drivers and computing power behind the sensor.

There is a large portion of time that the entire sensor is exposing. Once this cycle is complete (set by your shutter speed, aka shutter angle) then the reset cycle begins, which scans off the sensor in the same fashion. There is a short step where the black point is set here in the cycle as the levels are all set to off and then the cycle can repeat. The length of time for the sensor to be off is a function of the amount of time you wished the cycle to be open during exposure. The read/reset part of the time is extremely small.

I don't know the cycle time in the RED One but I do understand that it is pretty small. The Phantom HD has an incredibly small read/reset cycle time but then it has to because it is designed to be such a high speed camera that this is a necessity of function. It's just a few microseconds and it is impossible to ever see skew.

 

[Lawrence Bansbach]

I little bit of reading between the lines has to been done here, so I might not be right. Jarred said that the CF-Cards won't do REDCODE 100 and that REDCODE 100 would be "useful" for overcranking. So it might be the case that to achieve full overcranking you would need the SSD module. Has to verified though.

I believe it's been mentioned elsewhere that RedCode 100 would be useful for shooting at the highest frame rates with the least compression, but that shooting 120 fps at a lower data rate, such as RedCode 42, would still be possible. What's odd is that the original nonmodular design was projected to support RedCode rates up to 100 MBps even though the only storage medium ever mentioned for it was CF cards.

 

[Steven Caesare]

I think you still may be confusing sensor read time with shutter speed.

On a film camera the 4 microseconds refers to the time it takes for the leading edge of the shutter to move from the top of the frame to the bottom. If the film is 21.95mm × 16.00mm, 4 microsecs refers to the time it takes the leading edge of the shutter to travel 16mm.

I understand that in a still frame camera, but isn't Jim comparing this numbers to motion-film camera?

Unless I'm wrong, there is not a sperate high-speed sutter that has a shutter time in the single-digit microsecond range?

I've also found some places on the web guestimating some of the vDSLR's out now having sensor read times in the 25ms range... that would make the Scarlet/Epic target of 5ms much more likely... as good as Red's designs are, I don't think their sensors are 5,000 times faster.

Also this paper estimates a single CMOS sensor line read time at a coulpe of hundred microseconds, and we are talking a 2000+ lines to read on an Epic: http://docs.google.com/viewer?a=v&q...&gl=us&sig=AHIEtbSNqROi02cHt_dPSv8AqQ--wycrUQ

-sc

 

[Steven Caesare]

Steven, I saw the post where you say Jim referred to milliseconds. I'll try to find that. Mitch Gross also referred to skew in microseconds, however:



Here is another post from Mitch referring to microseconds:

I think he may not be incredibly clear, and confusing a line read time vs. that of the whole sensor, or getting his units mixed up.

But notice what he says at the end... the Phantom is a few microsceconds where it's impossible to see skew... if so and the Red One and now Epic are 9 and 5 microseconds than Skew would also be impossible to see.

Finally, here's a post on AbleCine.com (related to Mitch's company?), that refutes Mitch's assertion that the Phantom does it in a microscecond , and says that that it takes 1 millisecond- http://blog.abelcine.com/2009/04/15/the-difference-between-a-global-shutter-and-a-rolling-shutter/ :

And all rolling shutters are not alike. For instance, because of the high-speed nature of the Phantom cameras, the read/write time of the rolling shutter of a Phantom HD is just one millisecond.

Here are some other CMOS shutters for reference:

PixelLink's "fast reset" shutter takes ~200ms (http://www.pixelink.com/support/oem/cameras/common/image_capture/fast_reset_shutter.htm)

3 Megapixel CMOS takes 83ms (http://www.freepatentsonline.com/EP1933253.html)

Epix Inc industrial camera taked 16-33ms (http://www.epixinc.com/products/sv035.htm))

EZine article discusses 82ms for 3 mpixel sensor (http://ezinearticles.com/?Electronic-Shutter-in-CMOS-Image-Sensors&id=351614)


Now don't get me wrong, in relation to some of these designs, the Epic's read-reset speed of 5ms is easily 6-40x faster, which is amazing.

But I don't think Red has managed to make a 13 megapixel sensor that is 6-40 thousand times faster than some of these 3 megapixel units. Yet.

-sc

 

[Justin O'Neill]

Good finds Steven. I would love Jim, Jarred and Mitch to clear this up because I have been quoting them left and right!

 

[Steven Caesare]

Good finds Steven. I would love Jim, Jarred and Mitch to clear this up because I have been quoting them left and right!

I agree.

And FWIW, my intention was not to get in to a pi$$ing match here, but just to try and see to it the info is accurate... your thread is a great resource here and I appreciate it.

I hope somebody authoritative can chime in here and confirm for us all.

Thanks Justin.

-sc

 

[A83]

Upgrade sensor of 2/3" fixed?

Upgrade sensor of 2/3" fixed?

Will it be possible to upgrade the sensor of the 2/3" fixed scarlet in the future? (And throw out both old sensor and lens). Or is the interchangable the way to go if you want to upgrade later?

 

[Joseph Ward]

Will it be possible to upgrade the sensor of the 2/3" fixed scarlet in the future? (And throw out both old sensor and lens). Or is the interchangable the way to go if you want to upgrade later?

The new Red DSMC Brains won't be upgradeable.

 

[Joseph Hutson]

Will it be possible to upgrade the sensor of the 2/3" fixed scarlet in the future? (And throw out both old sensor and lens). Or is the interchangable the way to go if you want to upgrade later?

The new Red DSMC Brains won't be upgradeable.

Well, at least you won't get your money back from RED if you upgrade. However, any brain you buy will be fully functional with the modules you purchase presently/in the future.

 

[Bastien Tribalat]

There is something we do'nt know yet (I think) : will the SCARLET's body feature any kind of weather sealing ?

 

[Joseph Hutson]

There is something we do'nt know yet (I think) : will the SCARLET's body feature any kind of weather sealing ?

No word on this, but expect to see several 3rd parties wanting to create necessary accessories such as this... :thumbup1: