Incredible RED sensor

[wolf33d]

http://www.sonyalpharumors.com/red-just-dethroned-sony-a7rii-dxomark/

The new APSC 35mpx Red sensor beats everything in the world of stills camera. Higher dynamic range than D810 (15.2 stops!!!!) and incredible tonal range and high ISO

I do not understand how an APSC sensor like this can get that much DR and high ISO capability VS lets say a D810 sensor. Are RED sensor that much more advanced? Why dont we have APSC sensors with 15.2 of DR in stills camera? Or at leas in FF sensors. I dont get it.

If RED can do this why not Canon Sony Nikon or anyone else. 

I want a 35mpx FF sensor with those performance in DR and high ISO in a Canon mirrorless body compatible with Canon lenses with IBIS, 4K60p, FHD240p, Canon DPAF,  H265 150mbps codec + prores option, 10 bit. for 3000$. 

This will come in about 3 years. But damn... It exists today so why not

[Mattias Burling]

Before the A7s the best iso was an APS-C. It comes and goes and crop factor doesn't affect exposure. A 1" Rx100 kicks the ass of many older APS-C etc.

[Jimmy]

11 minutes ago, wolf33d said:

Why dont we have APSC sensors with 15.2 of DR in stills camera?

-----------------------------------------------------------------------------

I want a 35mpx FF sensor with those performance in DR and high ISO in a Canon mirrorless body compatible with Canon lenses with IBIS, 4K60p, FHD240p, Canon DPAF,  H265 150mbps codec + prores option, 10 bit. for 3000$.

I think you just answered your own question. That red sensor alone is probably costing the buyer $3k+

[wolf33d]

1 minute ago, Jimmy said:

I think you just answered your own question. That red sensor alone is probably costing the buyer $3k+

I dont think so. A sensor is a sensor. Sensor cost comes mainly from its size (cost of sensor raise with the square of the size) so I think a FF sensor cost more than this one. 

From comments in the link above someone says it is because of the treatment to RAW they bring, sort of temporal noise reduction combining multiple exposure in a pixel to reduce noise improving DR and high ISO, all in a 16bit package. But then again, why not do that in our DSLRs.....

[tomekk]

10 minutes ago, wolf33d said:

http://www.sonyalpharumors.com/red-just-dethroned-sony-a7rii-dxomark/

The new APSC 35mpx Red sensor beats everything in the world of stills camera. Higher dynamic range than D810 (15.2 stops!!!!) and incredible tonal range and high ISO

I do not understand how an APSC sensor like this can get that much DR and high ISO capability VS lets say a D810 sensor. Are RED sensor that much more advanced? Why dont we have APSC sensors with 15.2 of DR in stills camera? Or at leas in FF sensors. I dont get it.

If RED can do this why not Canon Sony Nikon or anyone else. 

I want a 35mpx FF sensor with those performance in DR and high ISO in a Canon mirrorless body compatible with Canon lenses with IBIS, 4K60p, FHD240p, Canon DPAF,  H265 150mbps codec + prores option, 10 bit. for 3000$. 

This will come in about 3 years. But damn... It exists today so why not

Red could be using some form of noise reduction and applying it to RAW which other manufactures don't do (yet:(). If true, ISO performance won't be that impressive. We need more tests ;). 

[Neumann Films]

Depends on which OLPF you use but with the "low light OLPF" I'm comfortable shooting ISO3200 on this sensor.

[noone]

57 minutes ago, wolf33d said:

http://www.sonyalpharumors.com/red-just-dethroned-sony-a7rii-dxomark/

 

This will come in about 3 years. But damn... It exists today so why not

Three YEARS?       What do you think Sony, Canon and others who make sensors will have by then?

Sony should be at the A7Riv and A7siii or iv by then!     Actually, my guess is A7Riii with the iv just announced and people whinging about their A7Riii being discontinued so quickly.         A7s iii will be popular and the iv will come a bit later in the year.

A9 will still be coming soon!

Then again, my crystal ball is a bit cracked and is a first version Sony anyway.

[tomekk]

53 minutes ago, wolf33d said:

From comments in the link above someone says it is because of the treatment to RAW they bring, sort of temporal noise reduction combining multiple exposure in a pixel to reduce noise improving DR and high ISO, all in a 16bit package. But then again, why not do that in our DSLRs.....

If they do it digitally, then it could as well be done in post leaving the choice to user I believe. Why would you want noise reduction applied on your DSLR without having any control on how it was applied? There are problems with noise reduction algos that could sometimes degrade an image. If it's applied without user's knowledge, it's cheating too. It's not a RAW sensor output anymore it's RAW + an algo output which makes the whole test skewed. Now, if you would want to compare other sensors to this one, you would have to denoise their outputs first and then compare. 
I'm not saying that's the case here, obviously.

[aldolega]

The argument for in-camera NR is that it is working with the raw sensor data, whereas with NR in post you're working with the compressed codec (on non-raw-video cams of course) and thus much less data. 

[Don Kotlos]

42 minutes ago, tomekk said:

If they do it digitally, then it could as well be done in post leaving the choice to user I believe. Why would you want noise reduction applied on your DSLR without having any control on how it was applied?

Every RAW output has some kind of preprocessing. The same way that hot pixels are specific to the sensor and are remapped in camera, noise can have a pattern that is sensor-specific and then is preferable to have it done in camera than having to do it every single time in post. Moreover if the camera doesn't record high bit depth RAW video, then you will get much better results doing in camera than in post. Now, of course if overdone then it will look like shit and if you don't have the RAW output you can't do anything about it in post  

42 minutes ago, tomekk said:

 If it's applied without user's knowledge, it's cheating too. 

I would use "mislead" instead but I get the point. For example Fuji's claims about having FF noise levels with Xtrans rely on 1.Noise reduction even in RAW files 2.Having different ISO standards than anyone else, so ISO 6400 is more like 3200 and 3.More green photodetectors. Great colors though! 

1 hour ago, wolf33d said:

I dont think so. A sensor is a sensor. Sensor cost comes mainly from its size (cost of sensor raise with the square of the size) so I think a FF sensor cost more than this one. 

Size is a major part of the cost, but pixel density, custom design, and production volume play a big role as well. Also these are exponentially increasing costs so its hard to estimate the real cost of such a sensor based on just the size. 

[Andrew Reid]

DXOMark have some questions to answer...

How can they put a score out for a sensor which applies high ISO noise reduction to the RAW files and benchmark it against sensors which don't do noise reduction on the raw files?

I seriously doubt the RED's Super 35mm chip with those densely packed small pixels beats an A7S II for noise.

2 hours ago, wolf33d said:

I do not understand how an APSC sensor like this can get that much DR and high ISO capability VS lets say a D810 sensor. Are RED sensor that much more advanced? Why dont we have APSC sensors with 15.2 of DR in stills camera? Or at leas in FF sensors. I dont get it.

The DR score also comes from heavy processing.

The sensor probably has twin A/D converters and other tricks going on for dynamic range.

The back end processing is necessary as well.

[Robin Billingham]

Just had a little look over at reduser to see more about the sensor and Jarred has posted that no noise reduction is taking place ..

he says 'but if course you already knew that' or something along those lines 

[tomekk]

1 hour ago, Don Kotlos said:

Every RAW output has some kind of preprocessing. The same way that hot pixels are specific to the sensor and are remapped in camera, noise can have a pattern that is sensor-specific and then is preferable to have it done in camera than having to do it every single time in post. Moreover if the camera doesn't record high bit depth RAW video, then you will get much better results doing in camera than in post. Now, of course if overdone then it will look like shit and if you don't have the RAW output you can't do anything about it in post  

I would use "mislead" instead but I get the point. For example Fuji's claims about having FF noise levels with Xtrans rely on 1.Noise reduction even in RAW files 2.Having different ISO standards than anyone else, so ISO 6400 is more like 3200 and 3.More green photodetectors. Great colors though! 

 

I used "cheating" because I'm tired of a political correct vocabulary. If we let companies off the hook in these kind of situations we'll get exploited until we die and it's just going to get worse, IMHO. I don't have anything against noise reduction or anything they do anywhere in the processing pipeline as long as they don't lie/cover it up so that objective tests can be done against those products. Generally, everything that's done digitally can be done with better results in post due to high processing power that's available and with much more precise control. Could you link me to an article about sensor-specific noise that needs to be dealt with in camera? Simple logic tells me that if there is a sensor-specific noise on top of a regular noise then it's a crappy sensor, but I'm probably missing something.  

[Don Kotlos]

17 minutes ago, tomekk said:

Generally, everything that's done digitally can be done with better results in post due to high processing power that's available and with much more precise control.

That would be true if you had access to the same digital data but especially in consumer cameras you get a heavily processed version of the original.

17 minutes ago, tomekk said:

Could you link me to an article about sensor-specific noise that needs to be dealt with in camera? Simple logic tells me that if there is a sensor-specific noise on top of regular noise then it's a crappy sensor, but I'm probably missing something.  

There are plenty of engineering articles but this answer at stackexchange provides a nice summary:

http://photo.stackexchange.com/questions/27318/how-does-in-camera-noise-reduction-compare-to-software-one

[tomekk]

1 hour ago, Don Kotlos said:

There are plenty of engineering articles but this answer at stackexchange provides a nice summary:

http://photo.stackexchange.com/questions/27318/how-does-in-camera-noise-reduction-compare-to-software-one

This is different to what I meant. All this time I meant only software noise reduction (my fault if I wasn't clear about it but I was relating to the article all this time and their findings). If it's hardware based then it's fine because it's something entirely different and cannot be replicated in software. 

1 hour ago, Don Kotlos said:

That would be true if you had access to the same digital data but especially in consumer cameras you get a heavily processed version of the original.

I'm not sure what you mean. Even in consumer cameras there is usually an option to adjust settings. Do you mean a consumer camera (or any camera really) would do a better denoising if I left in camera denoising option on (even if it's a JPEG) than I would do in Photoshop if I left it off?

[wolf33d]

I do not care that RED applies what ever correction and magic to their file and cook them for 10h before delivering it to my memory card. 

I care that what I get on my memory card is an absolutely fantastic looking 16 bit RAW image with 15.2DR and no noise. 
In a SS35 sensor... so imagine in FF..

I care that what we can buy for 3K today are 11DR in a GH5, or 14.8DR for stills in a D810 but with a body and logic (mirror) as old as my father and video capabilities of a 2010 phone. 

There is something wrong in this world. Why is there no US/EUR company making cameras? We have Apple in the US who completely changed the computer industry, then the music industry, now the phone industry and all those asian company copying what they do, taking them years to equal the product each time. 

Now why dont we have a Canon or Nikon from US? With people who can actually innovate and give us the best tech and innovations in products dedicated to photo and video. 
I am so tired of those Japanese company as classic as it gets serving us the same shit for years. Those guys are the most conservative people on the planet. Makes me sick. 

Now what a surprise RED is from California, and gives us a way better S35 sensor than any FF sensor ever done by whoever else. 
 

[Don Kotlos]

17 minutes ago, tomekk said:

Do you mean a consumer camera (or any camera really) would do a better denoising if I left in camera denoising option on (even if it's a JPEG) than I would do in Photoshop if I left it off?

For a compressed output data stream, in principle yes. Not only because noise estimation (and thus reduction) will be better with more data, but also because the compression algorithms handle noise very badly. For example: use an original file to export one H264 without any noise and one with noise applied at the same bitrate. Second file will look far worse.

Now that is not always the case. Denoising algorithms do depend on processing power and improve over time. So with RAW files or files with little compression, in post you can get much better results that in camera. In that case internal reduction of sensor specific noise (such as hot pixels/ hue casts/ QE inhomogeneity/ non-bayer sensor design, etc) is a matter of convenience. 

[Don Kotlos]

43 minutes ago, wolf33d said:

I am so tired of those Japanese company as classic as it gets serving us the same shit for years. Those guys are the most conservative people on the planet. Makes me sick. 

Arguing that Japanese are not innovative, thats a first. 

Quote

There is something wrong in this world 

I would say feeling sick over where an 8K camera is produced, but hey that is just my opinion. 

[noone]

1 hour ago, wolf33d said:

I do not care that RED applies what ever correction and magic to their file and cook them for 10h before delivering it to my memory card. 

I care that what I get on my memory card is an absolutely fantastic looking 16 bit RAW image with 15.2DR and no noise. 
In a SS35 sensor... so imagine in FF..

I care that what we can buy for 3K today are 11DR in a GH5, or 14.8DR for stills in a D810 but with a body and logic (mirror) as old as my father and video capabilities of a 2010 phone. 

There is something wrong in this world. Why is there no US/EUR company making cameras? We have Apple in the US who completely changed the computer industry, then the music industry, now the phone industry and all those asian company copying what they do, taking them years to equal the product each time. 

Now why dont we have a Canon or Nikon from US? With people who can actually innovate and give us the best tech and innovations in products dedicated to photo and video. 
I am so tired of those Japanese company as classic as it gets serving us the same shit for years. Those guys are the most conservative people on the planet. Makes me sick. 

Now what a surprise RED is from California, and gives us a way better S35 sensor than any FF sensor ever done by whoever else. 
 

Well, it seems  you can buy the sensor now in the Weapon 8k.       Cost $49,000 for the brain (EDIT $29,500 with a lesser brain).    Add a basic recommended package $5,643 (and you still can not just shoot).     That will take 2 to 3 weeks to ship from Red.

http://www.red.com/store/products/weapon-brain-with-helium-8k-s35-sensor

Those Japanese cameras will cost less than the BASIC package needed to be able to use it and you can walk into a shop in many places and buy one.

Many of the mass produced things are being made by people working for an awful lot less than if they were made in the US (or almost any western country).

I can see a point where there wont be any workers in factories (or very few) and at that point you might start seeing production coming back to Western countries as costs will be the same.      We might be there already for some things but not yet for mainstream cameras.     

THIS, is not a mainstream camera and I doubt there would be a huge number of employees in the RED factory/factories.

I hope that cameras ARE made in the west and soon though as long as I can get what I want and can afford (this isn't either- way beyond me) I don't care where it is made really.

[tomekk]

50 minutes ago, Don Kotlos said:

For a compressed output data stream, in principle yes. Not only because noise estimation (and thus reduction) will be better with more data, but also because the compression algorithms handle noise very badly. For example: use an original file to export one H264 without any noise and one with noise applied at the same bitrate. Second file will look far worse.

Now that is not always the case. Denoising algorithms do depend on processing power and improve over time. So with RAW files or files with little compression, in post you can get much better results that in camera. In that case internal reduction of sensor specific noise (such as hot pixels/ hue casts/ QE inhomogeneity/ non-bayer sensor design, etc) is a matter of convenience. 

Are you sure software noise reduction (the one you can turn on/off in camera options) happens on the RAW data and not on a compressed file? Playing with a JPEG file I'm pretty sure I can denoise it much better in Photoshsop than through in camera option... I'll check with H264 footage to confirm when I have some free time.