Is the RED Scarlet just pure hype?

[Deji Joseph]

First thing first, please this is NOT about RED bashing or film versus digital or which camera is better or the best, it is just about discussing the actual usability of the scarlet. The other day I came across so intriguing information as one does, in an internet article. It claimed in the days of 2/3inch betacam acquiring a lens which could resolve greater than a 1000 lines was apparently an expensive miracle. this still applies today with 1/2 inch cameras sharpening the hell out of the raw data from the sensor to make up numbers. this got me thinking, if optimo zooms for 35mm can resolve about 4k on a super 35 frame, what will the 'budget' scarlet fixed lens resolve? Arri says the theoretical effective resolution of S16 which is 30% larger should be 2058 × 1237 pixels, therefore isn't the scarlet massively oversampling? In my honest opinion I think it would have been smarter to give the scarlet 1080p, this means better noise to signal ratio and dynamic range. However since there is no physical camera to compare I might be wrong and will gladly accept the camera, what is your take on the scarlet?

 

Dj

[David Mullen ASC]

Remember that this is a Raw single-sensor camera, so measured resolution after conversion to RGB will be lower than the number of pixels on the sensor. A 3-sensor 2/3" video camera has a nearly 2K sensor per color, whereas the Scarlet has to generate 2K / 1080P for each color from a single sensor, so you want to oversample. Now they might have gotten away with a 2.5K sensor or so if the end goal was 2K / 1080P RGB finishing, but 3K is fine, especially if you want to use the 3K Scarlet to shoot B-roll for a 4K project.

 

Most decent lenses should resolve more lines per millimeter than the Scarlet sensor can capture.

 

You definitely wouldn't want to use a single 2K / 1080P sensor if you want to generate RGB at that resolution from it, though that's what the SI-2K did, which some people consider to have a soft-ish image as a result.

 

But as for making the 2/3" sensor 2.5K-ish instead of 3K, sure, you'd get a bit less noise, better sensitivity though also less resolution. I certainly wouldn't call shooting 3K Raw for finishing to 2K / 1080P RGB as "massive oversampling" more like a desirable amount of oversampling.

[Brian Drysdale]

The Scarlet is 3k beyer, which should allow the Scarlet to perform quite nicely for a 1080p finish. The 3k isn't the actual resolution of the Scarlet, just the pixel count, but on paper this should be just over 2k. No one outside RED has done a test, but people who have seen pictures to date are pretty positive. I'm not sure the Betacam lenses are really that impressive as an example, the old 16mm Cooke Varokinetal zoom was claimed to resolve 200 line pairs/mm and certainly the 2/3" video zoom they brought out was way more impressive than any other 2/3" standard def lens I saw.

 

If it's hype remains to be seen, but their later cameras seem to be performing pretty well.

[Deji Joseph]

The Scarlet is 3k beyer, which should allow the Scarlet to perform quite nicely for a 1080p finish. The 3k isn't the actual resolution of the Scarlet, just the pixel count, but on paper this should be just over 2k. No one outside RED has done a test, but people who have seen pictures to date are pretty positive. I'm not sure the Betacam lenses are really that impressive as an example, the old 16mm Cooke Varokinetal zoom was claimed to resolve 200 line pairs/mm and certainly the 2/3" video zoom they brought out was way more impressive than any other 2/3" standard def lens I saw.

 

If it's hype remains to be seen, but their later cameras seem to be performing pretty well.

 

Are "Digital zooms" softer because they are designed for 3 sensor cameras then?

[David Mullen ASC]

2/3" HD zooms are quite sharp, and expensive, just not the old 2/3" standard-def zooms, because they didn't need to be. But both types are designed to focus red, green, and blue wavelengths through a prism block to three different sensors, so they need an adaptor to get them to focus on a single sensor and I'm not sure if there is any sharpness lost in the adaptor - Mitch Gross could tell you since Abel Cine Tech has them.

[Deji Joseph]

2/3" HD zooms are quite sharp, and expensive, just not the old 2/3" standard-def zooms, because they didn't need to be. But both types are designed to focus red, green, and blue wavelengths through a prism block to three different sensors, so they need an adaptor to get them to focus on a single sensor and I'm not sure if there is any sharpness lost in the adaptor - Mitch Gross could tell you since Abel Cine Tech has them.

 

I agree, for the changeable Scarlet that is an option. The problem is at the price point of the Fixed scarlet, RED isn't going to be able to afford any decent glass. So wont the image be soft anyway? The question would be does a 3k output from a subpar lens offer any incentive over a 2k one?

[Brian Drysdale]

I agree, for the changeable Scarlet that is an option. The problem is at the price point of the Fixed scarlet, RED isn't going to be able to afford any decent glass. So wont the image be soft anyway? The question would be does a 3k output from a subpar lens offer any incentive over a 2k one?

 

It's not the glass that's usually the problem it's the mechanics. From the information the fixed zoom seems to operates more like a lens in a EX1 than a traditional cine lens, so if Sony can manage it on the EX1 there's no reason why RED can't on the Scarlet.

 

There is some test footage shot on the fixed Scarlet:

 

http://reduser.net/forum/showthread.php?59527-Scarlet-8x-internal-120FPS-test

[Brian Drysdale]

This was shot with the fixed Scarlet.

 

http://red.cachefly.net/rackfocusx.mov

[David Mullen ASC]

I don't get the logic that Red should make the Scarlet worse in resolution because some people won't be able to afford to put a good lens on it.

[Nicholas Bedford]

All I know is that I want one.

[Phil Rhodes]

Noise and sensor size are, in a strange way, related.

 

More pixels equals smaller pixels, which equals less incident photons, which means less signal, so more pixels on the same size sensor increases noise.

 

If the pixel is smaller, the noise artefact is less objectionable, especially if you scale down for distribution. This is the only way the first Red camera ever looked in any way acceptable - it was noisy as hell, a factor that I misinterpreted completely at first, but by the time you watched it on an HD display, you'd averaged about four pixels together and cut the apparent noise floor massively.

 

I wouldn't be surprised to find there was some underlying principle of physics that made apparent noise artefacts at a given observed picture size, for a given taking sensor size, more or less constant regardless of pixel density. Perhaps there is, perhaps we're just talking about compromises inherent to the photovoltaic effect.

 

P

[Keith Walters]

I don't get the logic that Red should make the Scarlet worse in resolution because some people won't be able to afford to put a good lens on it.

Yes, the vast majority of compact stills cameras sold these days have 10-12 Megapixel sensors, which in most cases greatly exceeds the resolution of the lenses fitted, but nobody is suggesting we should return to 3 megapixel sensors.

The reality is, the more you oversample, the less moire artefacts you get, and sensor pixels keep getting cheaper while quality Optical Low-Pass filters remain at about the same price.

[Deji Joseph]

I don't get the logic that Red should make the Scarlet worse in resolution because some people won't be able to afford to put a good lens on it.

 

I think I should note I'm mainly talking about the Fixed scarlet, which was designed for the lower end film maker. In my opinion I imagined this group wouldn't need more than fully resolved 1080p. Many filmmakers in this group love the DSLR because of its low light abilities. With such a High pixel density the Scarlet may not be able to compete. I think 1080p when resolved to its full potential can be adequately sharp, like comparing a 5D image to an F3 image recorded via external rec, one can tell the glaring difference in sharpness. I find it hard to believe can make a T2.6 28-224mm lens that will do 3k justice, but I am not an expert in Lens design or Lens for 2/3inch or super 16. The Scarlet at 8k costs as much as a Sony Ex1 right now and with the amount of power they are stuffing in the body Im skeptical they will have enough left to beat the Ex1 or Ex3s Lenses which arent 3k ready.

 

Dj

[Deji Joseph]

Yes, the vast majority of compact stills cameras sold these days have 10-12 Megapixel sensors, which in most cases greatly exceeds the resolution of the lenses fitted, but nobody is suggesting we should return to 3 megapixel sensors.

The reality is, the more you oversample, the less moire artefacts you get, and sensor pixels keep getting cheaper while quality Optical Low-Pass filters remain at about the same price.

 

But these are consumer products created via the "resolution wars". To the layman 12mp is better than 10mp, even though its more complex in reality. Ive met photographers who complain about the increase in noise due to less photons per pixel.

[Brian Drysdale]

The Scarlet at 8k costs as much as a Sony Ex1 right now and with the amount of power they are stuffing in the body Im skeptical they will have enough left to beat the Ex1 or Ex3s Lenses which arent 3k ready.

 

I wouldn't confuse the numbers, the 3k is a single beyer sensor (from which you extract 2k or 1,920 x 1080p), the EX1 has 3 x 1.920k (if you like) sensors. Being a 1/2" sensor, its lenses need to have a high resolution other wise the quality drops off and the Canon EF 300 needs to have an higher resolution lens to match the EX series on a 1/3" chip. The Scarlet lens seems to be within the ball park of these cameras, but with a smaller zoom range.

 

I imagine the Scarlet fixed will be used by quite a few documentary film makers.

 

The price of the Scarlet seems to be slowly drifting upwards, so the final sales price remains to be confirmed.

[Deji Joseph]

I wouldn't confuse the numbers, the 3k is a single beyer sensor (from which you extract 2k or 1,920 x 1080p), the EX1 has 3 x 1.920k (if you like) sensors. Being a 1/2" sensor, its lenses need to have a high resolution other wise the quality drops off and the Canon EF 300 needs to have an higher resolution lens to match the EX series on a 1/3" chip. The Scarlet lens seems to be within the ball park of these cameras, but with a smaller zoom range.

 

I imagine the Scarlet fixed will be used by quite a few documentary film makers.

 

The price of the Scarlet seems to be slowly drifting upwards, so the final sales price remains to be confirmed.

 

Your right, if its possible to do 1080p on 1/3inch the 3k on 2/3inch should be fine.

 

Dj

[John Sprung]

Your right, if its possible to do 1080p on 1/3inch the 3k on 2/3inch should be fine.

 

Dj

 

But is it possible to do anything at all well at the 1/3" level? Doesn't it crap out from diffraction at, IIRC, f/4?

 

 

 

 

-- J.S.

[Brian Drysdale]

But is it possible to do anything at all well at the 1/3" level? Doesn't it crap out from diffraction at, IIRC, f/4?

 

 

 

 

-- J.S.

 

On Alan Roberts' BBC test, the Canon XF 300 matches the EX cameras in resolution terms. He mentions its usable range being up to f6.8 and the diffraction kicking in after f8, qualifying that this is unusual for such a small sensor.

http://thebrownings.name/WHP034/pdf/WHP034-ADD50_rev1_Canon_XF300-305.pdf

[Nicholas Bedford]

On Alan Roberts' BBC test, the Canon XF 300 matches the EX cameras in resolution terms. He mentions its usable range being up to f6.8 and the diffraction kicking in after f8, qualifying that this is unusual for such a small sensor.

http://thebrownings....n_XF300-305.pdf

 

The problem with DSLRs right now is that most of them don't interpolate a final image, they skip lines to produce the image at the right resolution. This has the adverse effect of making the sharpness of the final frame being reliant on the focus sharpness at the sensor level, not the final image.

 

Another problem is that you get all that noise that crops up in individual pixels, not a sampled version of many pixels into the final frame's pixel. In my higher than 160 ISO tests, I'm finding it's not the easiest to get actual sharp results in full HD, especially when you're working at ISO 640.

 

Maybe I'm missing something?

[Brian Drysdale]

Maybe I'm missing something?

 

I'm not sure how this relates to the Canon XF300, since its a 1/3" HD camcorder.

Yes, the DSLRs have issues.