RED 8K S35

[Christoffer Glans]

I am perplexed as to why you keep saying this. From the beginning, Jarred said there would be a path to the next sensor for Weapon Mg. Period. When asked about a path to 8K VV, he said no, he did not consider there to be a path. Later he said something to the effect that his answer of "NO" was technically incorrect, because you could (at great cost) upgrade a Mg camera to CF, then go to 8K VV, but it wouldn't make any sense. So physically possible? Yes. Recommended? No. Why would he suddenly reverse course for you to get your Weapon Mg on Helium when he clearly stated there would be a "path". When he says path, I don't think he means upgrading to upgrade. YMMV.

Getting Weapon 6K for Epic MX users also had "a path", which were to upgrade to Dragon and then upgrade to Weapon. Having a "path" doesn't mean specifically that we will have a sensor upgrade like we got with the MX to Dragon upgrade. It could mean that the path need you to upgrade to CF first and then upgrade to Helium. Yes, we might get the s35mm 8K in the MG, but that could also mean higher compression and lower high speed capabilities and I'm not sure if loosing those aspects of the MG is worth getting 8K for. Only reason would be improvements on other areas of the sensor, such as DR, color, noise and so on.

Jared has stated that the Mg Weapon will be able to the the Helium. What he didn't say is if the Mg Weapon Helium would be 8k. It could be a 7k version for all we know at a slightly smaller sensor size. Or helium data is not as big, and therefore an 8k sensor would not need as powerful ASICS in the Helium environment. .

If Helium is 8K at s35mm it's a locked resolution to that sensor size in the same way as Weapon 6K has it's sensor size and to get 8K you need to increase the sensor size. Increasing the pixel density on the sensor to get the same sensor size as Weapon 6K but at 8K resolution means that to have the same sensor size on the Weapon 6K MG you will need to have 8K, there's no other choice. The manufacture of a sensor does not change pixel density since pixel density is part of what that sensor is. If you change the pixel density you will not have a Helium sensor anymore but something else and you need another line of production for that. So 8K s35mm is locked. If you mean to say that Weapon MG won't be getting 8K then you are talking about a lower resolution and sensor size like with the Raven.

Setting 8K aside, Helium on the MG still needs to be able to do up to 96 fps, have 4:1 normal compression and so on. Those things are more important than the resolution and if the price of 8K on the MG is to have less high speed capabilities and more compression, that is important to take into consideration.

 

[Rakesh Malik]

What I'm most excited for is to see how good the difference is in terms of IQ. It has always been a pretty decent jump Mysterium > MX > Dragon... I wonder if Helium will be much better in terms of DR, Noise, Low Light performance, etc. Although smaller pixels tend to mean less low light ability, I am hoping there are not just more pixels but better ones.

My guess on that front is that it's back-side illuminated.

 

[Milano Questa]

My guess on that front is that it's back-side illuminated.

It better be. 3.65microns is close to smartphone sensor territory. The Samsung NX1 has nearly the same BSI pixel size (3.6 microns). I guess that could give you an indication of how well Helium will perform in low light.

 

[Nils Ruinet]

Heh, I wonder how I missed all this information about Helium... :huh:

Well, as others, I'm pretty curious to see what improvements this new sensor will bring in terms of image quality.

The Dragon is already amazing in many aspects, DR, color rendition, resolution...

There is one point where I see some room for improvement : sensitivity. I'd really love to be able to shoot at 3200 ISO when needed.
That's why I'm a bit concerned when I hear that the pixels are now getting even smaller than before. I'd rather gets added sensitivity over more resolution right now. I mean, most producers are already giving me scared looks when they hear 6K. Most of the time I still end up shooting 5K, which is the sweet spot in my opinion.

Sure, 8K has its uses, but it will be a pretty long time before anything gets finished and shown in that resolution (appart from a few exceptions of course), and I'd rather see Red focus on other aspects than resolution for the near future.

But of course I know too little about this new sensor for now, and the other sensors they might be working on. So I'll just wait and see, in the meantime the 6K Dragon will still be able to produce gorgeous pictures, and that's not about to change anytime soon !

 

[Brandon Shafto]

I wonder when they're releasing an upgrade to the RED Rocket-X? Gotta have a new one for the Helium.

"The new RED Millennium Starglancer. One half the render times of Rocket-X at S35 in 8K. $7999."

 

[Antony Newman]

Nils,

The problem is Not (IMHO) one of sensitivity, but one of Signal-to-Noise ratio per unit Area of the sensor.

The Dragon sensor can already see into the Dark ... it is just that the chroma & luma are too noisy for it to be of usable.

Increasing resoluton should (even if this was smaller Dragon pixels) reduce our perception of solid blocks of noise - I would speculate that at some point our eyes perform a high-pass filter effect on the grain, and that we will find noisy images more pleasant.

AJ

...snip
There is one point where I see some room for improvement : sensitivity. I'd really love to be able to shoot at 3200 ISO when needed.

 

[Blair S. Paulsen]

(snip) The problem is Not (IMHO) one of sensitivity, but one of Signal-to-Noise ratio per unit Area of the sensor. (snip)
Increasing resoluton should (even if this was smaller Dragon pixels) reduce our perception of solid blocks of noise - I would speculate that at some point our eyes perform a high-pass filter effect on the grain, and that we will find noisy images more pleasant.
AJ

Agree, signal to noise ratio is the crux of the biscuit.

Also agree that noise perception does not always track linearly with technically measured noise levels.

Look forward to seeing side by side comparisons.

Cheers - #19

 

[Jarek Zabczynski]

I think the thought here is still that we're not in need of moving to CF. New sensor was always touted as "for Weapon / DSMC2 bodies" for the longest time. Also, with Epic-W being Helium... I'm not worried about the MG purchase choice. RED is moving quick, but they won't leave a new product behind like that.

I hear you, I'm just curious as to where the Epic-W will be placed in the foodchain, above or below our Mg Weapons.

 

[Bob Gundu]

I hear you, I'm just curious as to where the Epic-W will be placed in the foodchain, above or below our Mg Weapons.

It sits below the MG.

 

[Brody_B]

I hear you, I'm just curious as to where the Epic-W will be placed in the foodchain, above or below our Mg Weapons.

I think it is going to be right at the MG level. Then Red will discontinue MG and only offer Epic-W 8k S35. The MG upgrade to Helium S35 8K will essentially make it an Epic-W.

Bob how could Epic-W be any lower than MG, if MG is the lowest Brain that can handle 8K? Assuming that MG was earlier stated to not be eligible to upgrade to VV8K cause it wasn't good enough for the processing.

If course at this stage its all speculation, and some wild guesses. The fun part of this stage.

 

[Jarek Zabczynski]

I think it is going to be right at the MG level. Then Red will discontinue MG and only offer Epic-W 8k S35. The MG upgrade to Helium S35 8K will essentially make it an Epic-W.

Bob how could Epic-W be any lower than MG, if MG is the lowest Brain that can handle 8K? Assuming that MG was earlier stated to not be eligible to upgrade to VV8K cause it wasn't good enough for the processing.

If course at this stage its all speculation, and some wild guesses. The fun part of this stage.

It sits below the MG.

Epic-W will likely be Aluminum, and still 6k if I had to guess.

 

[Nick Morrison]

I guess that could give you an indication of how well Helium will perform in low light.

Not necesarrily. Those aren't 8K sensors. And they're also now powered by $20K boards.

 

[Nick Morrison]

I think it is going to be right at the MG level. Then Red will discontinue MG and only offer Epic-W 8k S35. The MG upgrade to Helium S35 8K will essentially make it an Epic-W.

Bob how could Epic-W be any lower than MG, if MG is the lowest Brain that can handle 8K? Assuming that MG was earlier stated to not be eligible to upgrade to VV8K cause it wasn't good enough for the processing.

If course at this stage its all speculation, and some wild guesses. The fun part of this stage.

I dont think Epic-W will be 8K. Helium yes. 8K no. My guess is 6K Helium - keeping the Epic-W at 6K, like its Epic Dragon brothers. Just like Scarlet-W is 5K ala its Scarlet Dragon bretheren. There is a logic here. There generally is.

6K Helium is approx Academy 35 fyi.

 

[PatrickFaith]

How does the defraction limit of s35 work when I have a 4K DI in something like prores4444 or openexr. So if I do T11 on a master prime at helium 8k but convert that in nuke to 4K openexr, will the diffraction limit not matter since my overall DI is at 4K ? What I think I'd want is a in camera wavelet transform so even if the sensor is 8k I'd get a 6k r3d but with a lot more dynamic range and higher ISO.

 

[Mike Nagel]

How does the defraction limit of s35 work when I have a 4K DI in something like prores4444 or openexr. So if I do T11 on a master prime at helium 8k but convert that in nuke to 4K openexr, will the diffraction limit not matter since my overall DI is at 4K ? What I think I'd want is a in camera wavelet transform so even if the sensor is 8k I'd get a 6k r3d but with a lot more dynamic range and higher ISO.

diffraction kicks in on sensor level and affects the capture... if u downscale, the algorithm will choose the better pixel out of 2 neighbors which may help but obviously as u know... there is no free lunch.

3.65 microns has a price.

will be interesting to see what real world results are...

 

[Brody_B]

I dont think Epic-W will be 8K. Helium yes. 8K no. My guess is 6K Helium - keeping the Epic-W at 6K, like its Epic Dragon brothers. Just like Scarlet-W is 5K ala its Scarlet Dragon bretheren. There is a logic here. There generally is.

6K Helium is approx Academy 35 fyi.

Interesting, I guess for some reason I just thought it would be Helium 8K and up. but I suppose there could be a 6K build. I guess I just assumed they wouldn't be going backward in resolution with the newest sensor, especially when its so small its intention is to maximize resolution in the limited footprint. So I get S35 footprint, but academy 35 or slightly smaller? I dunno, just doesn't seem right to me from a marketing standpoint. I guess we'll see...

 

[Lasse Mørkhagen]

If we get 12k sensor (helium vv?) it starts making sense to look into pixel binning.
This has probably been discussed before. But if you do a 4x4 pixelbinning on the 12k sensor you end up with a 3k output (if my basic mathskills have not failed me)

This gives you a much better noise level.... 3200 is suddenly as clean as 800.. or iso 800 as 200...
Phase One does this with their high-end digital cameras

If the binning is done at the sensor level you actually get a 3k raw file. as the sensor turns into a 3k sensor, but with much better specs

Interessting read,
http://www.phaseone.com/apsis/PhaseOneSensorplus.pdf

 

[Anthony Berenato Jr]

If we get 12k sensor (helium vv?) it starts making sense to look into pixel binning.
This has probably been discussed before. But if you do a 4x4 pixelbinning on the 12k sensor you end up with a 3k output (if my basic mathskills have not failed me)

To be honest, with a name like Helium - I'm honestly hoping something of the like is the secret sauce this go around. Being able to use the full sensor and still output a 4K or 6K raw file. At 8K, data is buck wild. Not all projects will need that much detail. Moving back down to 4K or 6K from a sensor the same-ish size as Dragon... We're running into some extreme crop problems that still run alongside data issues, mostly at 6K and moving down from there.

Some sort of method to "filling up and sucking air out of the balloon" would be absolutely fucking amazing from these cameras. I think that's a huge next step if accomplished.

 

[Milano Questa]

Not necesarrily. Those aren't 8K sensors. And they're also now powered by $20K boards.

I'm talking pixel performance. No downscaling.
Samsung NX1 probably has the best pixel / size performance ratio of the entire stills industry. I wouldn't be suprised if Samsung is producing the Helium sensors.

Boards can only do so much. A/D conversion might be slightly better. But Red has never been a low light champ. That's unlikely to change with Helium.
Or are you suggesting incamera noise reduction?

 

[Blair S. Paulsen]

Helium 6K is a possibility, but IMO, it's manifest destiny is S35-ish coverage 8K. It's the sweet spot.

While it's conjecture at this point as to whether there will be significant diffraction issues with a 3.65 micron pixel pitch, I'd wager that a purpose built OLPF will mitigate all but the most extreme instances. Obviously diffraction in the lens is still part of the equation, but that's true with 5 micron photo sites, film, ...

IMO, the more interesting issue is whether in camera downscaling from 8K to 4K will be too computationally expensive. Getting the image characteristics of an S35 format, with the Bayer penalty already accounted for, written to a media type that can be decoded in real time by modern CPUs and Pascal GPUs might really open up more use cases for the RED capture system. Interesting times.

Cheers - #19