Welcome to our community

Be a part of something great, join today!

  • 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 :)

No more FF35 Scarlet?

And what if Canon & Panasonic will have noSkipping models out next year?
5Dmk3 could have 24Mpx FF, but eg. 4/3" window (19.6 x 11 mm, 6 Mpx) for video without skipping. For 60fps this would mean only 360 Mpx/s. 7D already handles 144 Mpx/s and 1Dmk4 160 Mpx/s, so it wouldn't be too massive jump ahead.
At least Pana sees 4/3" as good option:
http://news.creativecow.net/story/863708
 
The sensor itself simply isn't powerful enough to output 5K+ RAW images 24 times a second. The 5D Mk. II can only manage 3.9 fps for 17 RAW frames. That's it. Even the 1D Mk. IV chokes out at 10 fps over 28 RAW frames. Doesn't matter if it outputs through HDMI or H.264. There's no option other than line-skipping; these sensors weren't designed for video.

Windowing to that extent means a complete loss of wide shots. It's not going to work.
 
The solution to me, which is what appears RED is doing, is as long as your sensor resolution is higher than your target, you can crop to any size you want or scale, since you are going from higher data to lower, there isn't a loss in quality. The less # of sensors they produce, the more time and effort can go into each one.

There's also a 3rd option, not to crop the sensor or do line skipping, which is what the Viper does. They call it DPM (dynamic pixel management). When shooting 720p, 6 photosites are combined for one pixel of output. When shooting 1080p, 4 are combined. When shooting cinemascope, 3 are combined.

No matter what mode you shoot, the output is massively oversampled and very detailed, as the sensor is not windowed for 720p and 1080p. Slightly windowed for cinemascope.
 
There's also a 3rd option, not to crop the sensor or do line skipping, which is what the Viper does. They call it DPM (dynamic pixel management). When shooting 720p, 6 photosites are combined for one pixel of output. When shooting 1080p, 4 are combined. When shooting cinemascope, 3 are combined.

No matter what mode you shoot, the output is massively oversampled and very detailed, as the sensor is not windowed for 720p and 1080p. Slightly windowed for cinemascope.

I am curious... by combining pixels like this, won't the image get much softer? Lacking sharpness?
 
There's also a 3rd option, not to crop the sensor or do line skipping, which is what the Viper does. They call it DPM (dynamic pixel management). When shooting 720p, 6 photosites are combined for one pixel of output. When shooting 1080p, 4 are combined. When shooting cinemascope, 3 are combined.

No matter what mode you shoot, the output is massively oversampled and very detailed, as the sensor is not windowed for 720p and 1080p. Slightly windowed for cinemascope.

Yes, but Viper sensor(s) has the power to process those 4 pixels to sample to 1080p. The DSLR sensor simply aren't powerful enough to do that. The only way they can do 1080p 24/30 times per second is by deactivating pixels altogether.
 
The sensor itself simply isn't powerful enough to output 5K+ RAW images 24 times a second.

I'm not extremely familiar with the engineering, but I don't think it is a sensor issue as much as it is post sensor processing, heat management, data throughput and read/write/storage issue. Is there someone in the house who actually knows?
 
I'm not extremely familiar with the engineering, but I don't think it is a sensor issue as much as it is post sensor processing, heat management, data throughput and read/write/storage issue. Is there someone in the house who actually knows?


It would be great if someone could confirm, but to the best of my understanding, it is a sensor limitation.

Edit: Here's a post from Jim: http://reduser.net/forum/showpost.php?p=581631&postcount=36. And once again: http://reduser.net/forum/showpost.php?p=581641&postcount=40

I think that pretty much confirms that it is indeed a sensor limitation. I don't blame you, it is a common misconception - an urban myth even - that these DSLRs could do 5K+ RAW if they had the post processing power. It is clear that the sensors themselves aren't up to the task.
 
Ah I see, thanks. However most processors are capable of greater speeds if sufficiently cooled. When Jim says the processor isn't fast enough, I still wonder if heat isn't part of the equation.
 
The sensor itself simply isn't powerful enough to output 5K+ RAW images 24 times a second. The 5D Mk. II can only manage 3.9 fps for 17 RAW frames. That's it. Even the 1D Mk. IV chokes out at 10 fps over 28 RAW frames. Doesn't matter if it outputs through HDMI or H.264. There's no option other than line-skipping; these sensors weren't designed for video.
Windowing to that extent means a complete loss of wide shots. It's not going to work.
5d2 was announced over 1.5 years ago. Canon's sensors and processors might have advanced since then. 1d4 can handle 160 Mpx/s, so it could handle 5Mpx with 30 fps. Window could be 2.5k x 2k. Or 2.9k x 1.7k (16:9) or 3.4k x 1.5 (2.39:1). If they would release a model (3d, 4d, 6d, or 8d?) specially for moving pictures, its sensor could have less pixels and it still would be pretty good for stills. How about 10Mpx sensor and 'scope window would be 87% of full sensors horizontal FOV? With FF sensor and 16:9 window with 16mm lens, it would give you 80 degrees. Wide enough?
 
It would be great if someone could confirm, but to the best of my understanding, it is a sensor limitation.
Edit: Here's a post from Jim: http://reduser.net/forum/showpost.php?p=581631&postcount=36. And once again: http://reduser.net/forum/showpost.php?p=581641&postcount=40
I think that pretty much confirms that it is indeed a sensor limitation.
But it might be able to output windowed 2.5k.
I think the limiting factor is not sensor or processor, but writing to CF.
How else it could be explaned that both 1d4 and 7d can shoot jpeg's 10/8 fps until the card is full?
http://www.dpreview.com/reviews/canoneos1dmarkIV/page13.asp
http://www.dpreview.com/reviews/canoneos7d/page13.asp
With jpeg there's even more work for processor than just moving raw data...
 
It would be great if someone could confirm, but to the best of my understanding, it is a sensor limitation.

Edit: Here's a post from Jim: http://reduser.net/forum/showpost.php?p=581631&postcount=36. And once again: http://reduser.net/forum/showpost.php?p=581641&postcount=40

I think that pretty much confirms that it is indeed a sensor limitation. I don't blame you, it is a common misconception - an urban myth even - that these DSLRs could do 5K+ RAW if they had the post processing power. It is clear that the sensors themselves aren't up to the task.

It's all of them. Sensor, heat (that produces noise in addition to just limiting the throughput), and processing power. It's a $2,000 still camera. And it produces wonderful video for a $2,000 still camera. But even if it could record raw to a $5000 add-on, how would that make sense? It makes sense because it's cheap, and looks good when used within its limits, but when you go outside those limits, it looks bad and costs too much.
 
I am curious... by combining pixels like this, won't the image get much softer? Lacking sharpness?

Quite the opposite Terry. You're thinking of this the wrong way. Don't get hung up on photosites vs output pixels. Ever taken a large image and scaled it down? Viewed 1:1 the scaled down image looks very sharp and antialiased right?

Now imagine you have an sensor that captures images FOUR times larger than 1080p and scales that captured image down to 1080p in real time. The downres makes very detailed 1080p sized images. More detailed than you can get from the same sensor design only capturing 1080p 1:1.
 
It's all of them. Sensor, heat (that produces noise in addition to just limiting the throughput), and processing power.
I still can't understand how you can claim all of them.
7D can push 143 megapixels per second from sensor, compress it to jpg and write 48 megabytes per second to a cf card.
What in these performance specs tells that it couldn't read 5Mpx window from sensor 30 times per second and compress that and write it to cf?
Windowing isn't implemented yet, but nothing indicates that it could not be done.
 
Mpx/sec is not the data rate. It is the data rate that determines processing power requirements.

5mp raw at 12bits/pixel for Scarlet is 1.8 gigabits per second data rate vs 2.2mp at 8 bits/pixel for Canon HD video = 528 megabits per second uncompressed. More than 3 times the processing power, more than 3 times the read/write speed required and Canon is very slow at this already. Power consumption for RED S35 Epic is somewhere around 60watts, vs around 12 watts for a typical HD camcorder and even less for a DSLR. You can't get something for nothing.

A 40 second 4k resolution .r3d clip at Redcode 36 that I have on my laptop is a 1.5 gigabyte file. 40 seconds of highest quality video from a 7D is approx. 220 megabytes.
 
Mpx/sec is not the data rate. It is the data rate that determines processing power requirements.
Thanks for reminding that. Although I believe that all of us in this forum knows this.
If you prefer data rates, 7D samples in 14 bits so that 143 Mpx turns to 2 Gb/s in adc.
5mp raw at 12bits/pixel for Scarlet is 1.8 gigabits per second data rate vs 2.2mp at 8 bits/pixel for Canon HD video = 528 megabits per second uncompressed. More than 3 times the processing power, more than 3 times the read/write speed required and Canon is very slow at this already. Power consumption for RED S35 Epic is somewhere around 60watts, vs around 12 watts for a typical HD camcorder and even less for a DSLR. You can't get something for nothing.
Okay, so because much more advanced camera (Epic) uses a lot of power, Canon can't implement windowed video recording to its next model? Logics?
7D does put that same 2 Gb/s through its pipes in still mode than Scarlet will in movie mode. Next Canon could do that also in movie mode. Power consumption is really just about compression. Canon doesn't have to choose as high quality than Red, so it might use less energy.
 
7d does not sample 14 bit for video and does not sample the full sensor for video. 90% of the sensor data is thrown out before A/D and encoding. The maximum full sensor raw frame rate for the camera is 8fps for up to 120 frames. Nowhere near the frame or continuous data rates necessary for motion.
 
Ok, since this question has been out there for a long time, here's the answer from my chat with Jim last night, to the best of my recollection...

We still have plans for an FF35 camera other than EPIC. We are not sure if it will be called the "Scarlet" or if it will be something else yet.
 
I wouldn't be at all suprised if he was to change the name. Scarlet has an image problem that may be impossible to shed, particullarly if the Scarlet audience has to wait another 1.5 or 2 years to get the promised FF35. Changing the name might help them put those things behind them... So long as they deliver the fixed by the Fall and the S35 before year's end.
 
Ok, since this question has been out there for a long time, here's the answer from my chat with Jim last night, to the best of my recollection...


They think it will take too much business from Epic. I guess two years is how long it takes a company to go from "the underdog that fights the big boys" into an "only bottom line" company. I don't have a problem with it, that's why people start businesses... Now we will also see more proprietary stuff, more regulation on this forum, prices will be "updated" north... etc.

And the funny part is that the only reason this happened is because Red One owners were complaining that they were "left out" etc etc... there was a lot of crying around here last year, so... kind of funny.
 
Back
Top