2.5K S35 Please

[brainburst]

"Strictly speaking, smaller pixels are higher technology just by definition, in the same way that 45 nm process is higher technology than 90 nm process."

I have never heard such utter rubbish. A higher technology process does not equate to a necessarily higher quality product. The higher tech process is meant first of all to yield more sensors per wafer. All these efforts are to squeeze as much performance as possible from as little silicon as possible. Apply identical principals to larger celled sensors and the performance yield is exponentially higher.
And you conveniently dodge the question. If two sensors use identical technology but one has larger photosites which one will perform better given the same amount of pixels? According to the paper you cited provided:

"pixel size reduction
makes sense as long as increased spatial resolution
can be achieved."

In essence, the rule only applies if one is increasing pixel density for an equivalent area. At equivalent pixel counts larger sized pixels are better!
Also noteworthy in the article you cited the picture from the 2.2 upixel looks better than the 1.75.

 

[Daniel Browning]

Great as the RED One may be, people still pine for:
- more dynamic range
- more low light sensitivity
- higher frame rates
- easier data wrangling

Agreed. These are helped by larger pixel sizes, but there might also be ways to get the same performance from smaller pixels. Frame rates, especially, might pose special difficulties for smaller pixels.

It may be possible to address some of the render/CPU time concerns with half-full and half-medium style de-Bayer/resample software that RED does; however, I don't think the quality is as good as the equivalent amount of CPU time spent on data from larger pixels. I could be wrong, but it seems like proportionately more CPU time must be spent in order to get equivalent performance from smaller pixels.

Storage concerns are easier. They might be addressed with higher compression ratios. For a given file size and viewing distance, higher resolution with higher compression ratio has much better quality than lower resolution with lower compression. So you can usually increase quality, or at the very least keep it the same, for a given storage size.

Dynamic range at high ISO has been improving faster than dynamic range at low ISO for DSLR in the last few years (ever since the 20D), even as the pixel sizes shrink, but it's possible that there is a limit for how small they can make the on-sensor CDS and other transistors until it can't make enough room for the photodiode to keep the same FWC per area. Dynamic range at low ISO, on the other hand, is much more favorable to small pixels.

 

[combatentropy]

Graehme said a long time ago that the sensors all resolve only about 75 (or 78 or or 73 or somesuch) percent of the actual pixels. So, if you do the math, you already have something in the neighborhood of 2.5k from a 3k sensor.

The goal is HD. Because of the very fact you cite, I said 2.5K.

2,500 * 0.78 = 1,950.

 

[Daniel Browning]

I have never heard such utter rubbish.

Please take care to avoid passionate overreactions; I am not interested in having a conversation with someone who is not respectful.

A higher technology process does not equate to a necessarily higher quality product.

I did not say that; I said "performance would be the same". When I said "Strictly speaking...", I was simply pointing out that one of the meanings of "technology level" is photolithography process. It could be said that a smaller process is "higher technology" even if the quality of the product is worse.

And you conveniently dodge the question.

I did answer it: I said performance would be the same.

If two sensors use identical technology but one has larger photosites which one will perform better given the same amount of pixels?

Combatentropy and I were discussing fixed sensor sizes; his question did not include "given the same amount of pixels". Your question compares a large sensor with a small sensor. The larger sensor performs better.

The cost of sensor, availability of lenses, etc. are reasons why I think it is much more useful to discuss fixed sensor sizes.

At equivalent pixel counts larger sized pixels are better!

I think it's better to say "bigger sensors are better". It has the exact same meaning, but is more direct.

Also noteworthy in the article you cited the picture from the 2.2 upixel looks better than the 1.75.

I cited it for the words and the men behind them, not the pictures in the PDF. Here is one with better pictures:

http://forums.dpreview.com/forums/read.asp?forum=1018&message=28607494

 

[brainburst]

Random comments

Random comments

Dynamic range at low ISO, on the other hand, is much more favorable to small pixels.

Dynamic range is more favorable to smaller pixels only when the ratio of pixels per area is fairly large.

On the other hand I simply don't get the wish for 2.5k. All of the problems cited with a slightly larger data size are well on the way to being solved. Storage is getting cheaper, computers & graphics adapters are getting faster. You don't develop products to solve soon-to-be yesterdays problems.

3k @ 4/3 is really the best compromise....

 

[Daniel Browning]

On the other hand I simply don't get the wish for 2.5k. All of the problems cited with a slightly larger data size are well on the way to being solved. Storage is getting cheaper, computers & graphics adapters are getting faster. You don't develop products to solve soon-to-be yesterdays problems.

Agreed. I wonder if any of the common post-processing tasks are fundamentally single-threaded by their nature. I know demosiac, curves, and resampling can scale well. Noise Ninja doesn't scale, but that might be due to something else.

 

[Katty Galloo]

Hi,

Pal has nothing to do with it, we are discussing an full HD camera with the EX! not an up resed SD camera like the HVX.
Remember the 100 data rate is based on 60fps, so at 24/25 it's nearer 40.
Looking at the final pictures is the best way to judge equipment IMO,

Stephen

Allwright, I got it , you guys seem to to a big fan of the EX and I am not in a mood to discus the technical difference becourse its not superior big
and like many others its not RAW full res either as you said. Indeed the Sony will have some advantage, It better be with his 2/3 sensor against the HVX with his 1/3sensors. As I said before, next year Panasonic and others will reply soon on the EX series and before you know it, the EX will be history again just like the HVX is now allready.

This proves again that a bigger sensor has more advantage.
That is why I vote for a scarlet with at least 3K and 4k and a s35 sensor at a good price, to stay ahead ond the competition.

Euro girl, :calm:

 

[Antoine Baumann]

If two sensors use identical technology but one has larger photosites which one will perform better given the same amount of pixels?

yeah but this will lead that one sensor is bigger than the other, isn'it.

I would rather have a S35 sensor with 4K resolution compare to the same S35 sensor but with 5K. The 4K would have bigger photosites than the 5K, but which one would be better?

BTW, related to the topic, I personnaly have the feeling that 2.5K is not enough resolution for a S35 sensor.

Thanks,
antoine.

PS: sorry I haven't got the time to read the documentation provided in this thread, but will surely do it.

 

[brainburst]

That is why I vote for a scarlet with at least 3K and 4k and a s35 sensor at a good price, to stay ahead and the competition.

Again I have to say I grow weary of people asking RED for the impractical to financially impossible. There is simply no compelling reason from Red's perspective to make a s35 sensor @ only 3k. It will require reengineering a totally new and incompatible wafer from their other s 35 sized sensors. It will cost the same to manufacture, more if you include the re-engineering costs. The only way it makes sense is if you are willing to pay the same amount or more for the 3k as for the 5k.
A 4/3 sized 3k sensor on the other hand will give you 99% of what you want. Larger pixels, better DOF. It's 1:1 compatible with 4/3rd lenses & 2:1 compatible with FF lenses. More importantly it can be made using the same fabs as across all of their other sensors. It thus represents a potential profit point for Red since they can yield more 3k sensors per wafer.

 

[Katty Galloo]

Again I have to say I grow weary of people asking RED for the impractical to financially impossible. There is simply no compelling reason from Red's perspective to make a s35 sensor @ only 3k. It will require reengineering a totally new and incompatible wafer from their other s 35 sized sensors. It will cost the same to manufacture, more if you include the re-engineering costs. The only way it makes sense is if you are willing to pay the same amount or more for the 3k as for the 5k.
A 4/3 sized 3k sensor on the other hand will give you 99% of what you want. Larger pixels, better DOF. It's 1:1 compatible with 4/3rd lenses & 2:1 compatible with FF lenses. More importantly it can be made using the same fabs as across all of their other sensors. It thus represents a potential profit point for Red since they can yield more 3k sensors per wafer.

I dont know about that statment brainburst. There are DSLR's on the market with an allmost full frame CMOS 22,2 x 14,8 mm, 3888 x 2592 pixels like Canon EOS 1000D and Nikon F60 for less then 1000$ how do you explain that! and there are other brands even cheaper, and they are also high quality sensors. I believe within 2 years there will be a lot of change!
RED has opened pandoras box now for a new age.

 

[proaudio4]

As I said before, next year Panasonic and others will reply soon on the EX series and before you know it, the EX will be history again just like the HVX is now allready.

I doubt it will be that far of a leap. Panasonic has stuck to their lower rez CCD and chose pixel shift for a while and have not budged. I really thought they were going to throw something on the table at the last show, but they did not.

Also, I'm willing to bet Sony is going to offer the next EX cameras with 4:2:2 @ 50mbps. There's been talk.. If they were smart, they would jump right over 50mbps and offer their 4:2:2 100mbps long-GOP which actually looks really good. Convergent Design XDR is using this as one of their codecs.

But having said that, RED Scarlet is giving them a run for their money... to say the least! RED is listening to their customers. Big companies like Sony and Panasonic like to spoon feed their customers on what they feel we should be using.

 

[brainburst]

I dont know about that statment brainburst. There are DSLR's on the market with an allmost full frame CMOS 22,2 x 14,8 mm, 3888 x 2592 pixels like Canon EOS 1000D and Nikon F60 for less then 1000$ how do you explain that!

Yes but the frame rate on those sensors is lacking.

You are clearly not understanding what I said. Do you know what it costs to come up with a completely different sensor design? Plus on going production costs? All of Reds cameras with the exception of the 2/3" can use either the Monstro or Mysterium x fabs across the entire product line. They can simply slice or dice the wafer to make a different sensor size. Get It? They can't do this with a 2.5k or 3k sensor at full frame size! It would have to be a completely different sensor design!
On the other hand the 4/3 size could be just another chip off the other wafers. It would fit right in with the production line.
I have a feeling if Red doesn't do something in this space some other company will seize the opportunity.

 

[Katty Galloo]

Why would you want a 4/3 chip If you can buy an S35?, I mean the 2/3 Scarlet will do exactly what you want even as a Pro. And be honest, what in the hell are we complaining about here. The Scarlet will have even better quality than the Viper and Sony F35 witch has been used for great cinema film productions! and all of this for what price? , my God can you feel the impact already? If you go by to what I have just said. Even I just start to realize we cant complain at all, its phenomenal!

Euro Girl :waaa:

 

[brainburst]

The Scarlet will have even better quality than the Viper and Sony F35

UHHH NO. Those are 3 CCD cameras. CCDs in general are better than CMOS. The Pixel sites are larger and the full resolution of the sensor is realized.

The 2/3" Scarlet is a bayer sensor so you don't really get 3k res. The pixel sites are smaller.

Scarlet will be competitive, but without great glass it won't be as good.

4/3 would have been better because of the larger sensor size, ability to use FF Glass and better dynamic range.

 

[Katty Galloo]

UHHH NO. Those are 3 CCD cameras. CCDs in general are better than CMOS. The Pixel sites are larger and the full resolution of the sensor is realized.

The 2/3" Scarlet is a bayer sensor so you don't really get 3k res. The pixel sites are smaller.

Scarlet will be competitive, but without great glass it won't be as good.

4/3 would have been better because of the larger sensor size, ability to use FF Glass and better dynamic range.

Ok, I agree about the glass, but you can use FF on the s35 wright.
and the CCD chips suffer from highlight blow outs and harder collors.
I'am all for CMOS. I have seen the movie Miami Vice shot with the Viper
and the image was quit grainy and dark, actualy ugly to me.
Lets say they have both pros and contras but thy are defently worth one and other.

see this link:

http://www.dalsa.com/markets/ccd_vs_cmos.asp


Euro girl

 

[REDneck]

having read that article on CMOS vs CCD, I'm now curious to know the story of RED's battle with the deficiencies of CMOS, and why it chose CMOS in the first place. it looks like it's the real underdog of sensors.

 

[David Mullen ASC]

ARRI chose CMOS as well for the D20/21... it partly may be due to size, apparently the separate A/D processor needed for a CCD requires a little more physical space, it may also have other requirements (in terms of power, temperature, processing time).

 

[Roberto Lequeux]

I am sure they have carefully weighed the options and arrived to the option that will yield the best possible quality and performance. Also Mr. J has proven excellent at thinking outside the box, so I feel secure about this one.

100% best quality possible 2k output (or FF1080) is what most need.

 

[Daniel Browning]

UHHH NO. Those are 3 CCD cameras. CCDs in general are better than CMOS. The Pixel sites are larger and the full resolution of the sensor is realized.

Brainburst, there are mistakes in your sentance due to the conflation of two separate issues: 1. CCD vs CMOS and 2. 3-chip vs. Bayer.

Lets say they have both pros and contras but thy are defently worth one and other.

http://www.dalsa.com/markets/ccd_vs_cmos.asp

having read that article on CMOS vs CCD, I'm now curious to know the story of RED's battle with the deficiencies of CMOS, and why it chose CMOS in the first place. it looks like it's the real underdog of sensors.

That "article" is heavy in opinion and weak in evidence, but that's what I would expect from a company so heavily invested in CCD technology.

CCD has been in development since the early 1970's, and when CMOS was brand new in the 90's, so the article was true about a decade ago. However, those issues have been long solved by clever sensor designers. Despite surpassing CCD in most markets, CMOS technology still shows no sign of plateu.

 

[Katty Galloo]

Hey,

I have some experience over the years with CCD and sinds I have seen
the Red data I am convinced this is the picture who comes close to the
film look qua collors and contrasts. CCD might be good over the years
but I think CMOS has took a giant jump upwards.
So I believe CCD will remain the choice for broadcast on TV
But when it comes to digital film in compared with the standard film
CMOS will not be beated. And there are other factors involved due this
like low pass filters etc.. Its a matter of choices I gues.

Euro girl