RED 222 razorsharp on 2K 120p

[Daniel Reichenbach]

But when someone says there is no difference then on a purely technical and informational standpoint I must take issue with this.

Hi Mitch, don't hope, that you talk about me. :biggrin: By the way, I rented your Phantom to shoot this great stuff, perhaps you don't know...., and it is great stuff, as you can see here: http://elementp.ch/de/portfolio/commercial/bernaqua.php
I know, technically, there is a difference, but sometimes there is to much discussion about technique and at the end, the result in an artistic way, is rather poor. I felt, that RED and Phantom played well together, that's all I wanted to say, and my feeling is, that I can't see a really difference between both pictures. It's of course subjectively and not a technical statement. But really, I don't care this much about technique if it works for me and gives me the picture I wanted and imagined.

 

[Mitch Gross]

Daniel, fair enough. If it works for you than go for it. Lovely spot, by the way.

 

[Daniel Reichenbach]

Thanks, this are just some picture put together, the real spot will be finished with more pictures and a kind of story in about a month ;-)

 

[Tom Lowe]

Mitch, thanks for the info. No need to apologize for posting the results of your tests; that is valuable information.

Regarding Red One, as you say, it only makes logical sense that a 2K crop of the sensor will be softer than a 4K downsample. That's just basic digital imaging 101.

But Phantom HD only has 2048 X 1108 active pixel Bayer-pattern CMOS chip, right? So wouldn't it be subject to the same type of "pixel for pixel" softness that Red One is when cropped at 2K?

 

[Deanan]

But Phantom HD only has 2048 X 1108 active pixel Bayer-pattern CMOS chip, right? So wouldn't it be subject to the same type of "pixel for pixel" softness that Red One is when cropped at 2K?

The major factors affecting the resolution between bayer patter cameras:

A. The spec of the OLPF and how much aliasing it cuts down
B. Pixel pitch in relation to the lens FOV and design
C. Debayer

Regarding C, if you compare 2k normal vs 2k high in RED rushes, you'll
notice that one appears sharper (normal) but at the expense of being a bit aliasy.

Regarding B, a larger pixel does not need as much resolving power but it does need a much stronger OLPF because the pixel size is greater.

 

[Mitch Gross]

But Phantom HD only has 2048 X 1108 active pixel Bayer-pattern CMOS chip, right? So wouldn't it be subject to the same type of "pixel for pixel" softness that Red One is when cropped at 2K?

The Phantom HD's sensor is 2048x2048 and since they are square pixels the math is simple for however one would shoose to crop them. The size of each of these pixels is more than four times the area of the RED One's pixels (12.5 micron square). This is why the 2K image area of the Phantom is bigger than Super-35 (26.5mm for 2K and S-35 is 24mm) while the full 4K of the RED One is about the same as Academy width (IIRC it's 22.2mm and Academy is 22mm; 4.5K on RED One would be about the same as S-35). Again, not a knock on RED, just a pile of numbers to answer a question. Phantom's OLPF needs to be stronger than RED One's to deal with the larger pixel size as Deanan noted. But there is a choice (okay, a vast series of choices) in deciding how strong one wishes to anti-alias using an OLPF and Graham has long stated his belief in a strong OLPF. I'm not questioning it at all, just noting that it is a stronger choice relative to pixel size (technically photosite size since we're getting down to fine points here) compared to Phantom's OLPF strength.

One can never assume that "all things are equal" mainly because they are not.

 

[Deanan]

OLPF strength is definitely a choice that is somewhat subjective. FWIW, we've tested stronger and weaker a couple of times and still ended up making the same decision each time.