Three three-letter cameras: EX1, F23, RED

[Álex Montoya]

And why's that? You supposedly have to show the cameras at their best.

 

[Dylan Reeve]

The fairest comparison you can do between the two is the following

Not really... Because footage shot on the F23 is very unlikely to be blown up to 4K. It may be scaled to 2K but given the similar size of 1080 and 2K it's likely that it'll remain native resolution. Similarly not a lot of RED footage will ultimately remain 4K, in many cases it will get a lot smaller, but 2K and 1080 are the most likely finishing resolutions for most things I suspect (they are the most practical).

The tests aren't really Apples to Apples, but it seems like a fairly good overview. But obviously there is room for improvement - it was more or less a test of the cameras as they are 'out of the box' but in practice in most serious situations they will all be tweaked to some degree. What I think was most impressive about that test was how well the EX1 performed. It really is a pretty amazing little camera.

And why's that? You supposedly have to show the cameras at their best.

Well indeed. And the F23's best isn't scaled up by a bit more than 200%. If you wanted to do a level-playing field test then it would be at the lowest common setting, as there's no inherent need for the other cameras to produce 4K images. So you'd do the tests at 1080, and scale RED's footage down to meet that mark.

 

[Álex Montoya]

The fairest comparison to get a sense of what each camera offers in terms of resolution is the one I did.

The RED just has more of it.

 

[Jeff Coatney]

It is interesting to note how the F23 evolved to emulate the characteristics of a film camera with it's film magazine blimp of a deck stuck on it. You can really see the wall the designers were up against in that development culture. To put so much focus on the form factor, as some kind of homage to film? Silly. The RED body is just so much more efficient and compact. That F23 is the first time I've seen NEW technology actually make a product BIGGER and more ungainly. How does that monster behave when it's on a technocrane with that camel's hump throwing the center-of-gravity off nodal axis?

 

[Dylan Reeve]

The fairest comparison to get a sense of what each camera offers in terms of resolution is the one I did.

The RED just has more of it.

Well that's clear just reading the sales documents, but that isn't all that matters.

I can take a Phantom camera that shoots 1000fps and compare it to a RED but the RED can only do 120fps, so I'll slow RED down in post to match the speed of the Phantom. Then I can say 'look the Phantom is much better, the RED footage is all jumpy and crap.' - You don't need to do any tests to know that Phantom shoots slower slow-motion than RED, and you don't need any tests to know that the RED shoots a bigger raster than the F23.

 

[Álex Montoya]

Well that's clear just reading the sales documents, but that isn't all that matters.

It may be clear to you, but many people are arguing after seeing those res charts that the RED is softer than the F23.

I just wanted to prove that it's not the case. The F23 even starts showing aliasing around the 900 figure while the RED holds up pretty well even in the 1200 and my guess is that with a better lens and a tighter f-stop it can do even better.

 

[Craig W. Bickerstaff]

It is interesting to note how the F23 evolved to emulate the characteristics of a film camera with it's film magazine blimp of a deck stuck on it. You can really see the wall the designers were up against in that development culture. To put so much focus on the form factor, as some kind of homage to film? Silly. The RED body is just so much more efficient and compact. That F23 is the first time I've seen NEW technology actually make a product BIGGER and more ungainly. How does that monster behave when it's on a technocrane with that camel's hump throwing the center-of-gravity off nodal axis?

This may shock and stun you but that concept was actually introduced by Panavision with the Genesis which is an extremely popular camera in the industry with praise on its design. Sony is simply doing what the industry is asking them to do. It's new tech influenced by older tech nothing wrong with that it doesn't make the red design any less relevant either.

 

[Andrew M.]

I looked at the Adam’s tests and one thing is missing, the information on how much of sharpening was applied to the material from EX1 and F23 cameras.
Also the scaling from 4K to 1080p is not just “I scaled it down to...) thing.

You can screw up all good in scaling and sharpening.
I can sharpen and scale picture to the point that every pixel in the RAW will be visible as a separate line, and what good it will produce?

I see perfectly smooth artifact and aliasing free picture samples compared to over-sharpened noisy pictures trying to estimate the vertical resolution of lp/ph

In terms of evaluating the pictures in this aspect we have to push both pictures to the point where artifacts are equal and then read the lp/ph out of it.

Can we get un-sharpen RAW from EX1 and F23 or it is already sharpen before RAW is produced?

 

[Graeme Nattress]

When I ask Jarred or Deanan to send me some test files in the email, or when I'm down at RED and Jim shoots me a test chart, the resulting R3D files are significantly sharper than those posted by Adam. I have lens test charts shot by other RED users, and they're all sharper than those posted by Adam.

When I take one of our 4k lens test charts I see practically perfect 2k zone plates, with full detail and contrast, right out to the edge, no aliasing to speak of and no sharpness added or needed. On a 4k zone, I see again practically perfect results out to 2k, still going strong at 2.5k, starting to fade at 3k, then gone by 3.3k. If I sharpen a bit (not excessive by any means) I can see detail beyond 3.2k, but I'd not count that. Beyond 3k there's some mild aliasing (which at least shows the lens used was sharp enough (Cooke 65mm). Our 18-85 is sharp enough to show aliasing out at 4k, again, showing that it's sharp enough too.

Andrew is right that saying "I scaled it down..." is not good enough. There are more ways to scale an image than I've had hot dinners....

Graeme

 

[Steve Sherrick]

I'm really curious about the way I'm seeing these debates carried out. It strikes me that a lot of people turn it into a personal thing. It's not just on this site and CML. I've seen it just about everywhere the RED camera is discussed. I suppose some people are just passionate about what they like and they aren't afraid to show it. But, the conversations I like tend to be the ones that look at an issue, discuss it in a professional way with examples and good context. I'm even okay when it gets a bit techy and as some would say, boring. The responses to Adam's article seemed to go down that road as agressively attacking RED.

I suppose with all of the hype, there was going to be some backlash but man, it gets pretty intense.

 

[Graeme Nattress]

Have you notices that most of the follow up posts to Adam's article follow the same theme, and have similar "generic" email addresses, that just happen to bounce?

I'm as passionate about education as much as I am about Red. That's why my posts are based upon scientific fact, and I try to teach when I write. You'll notice that you don't see that kind of posting from these trolls.

Graeme

 

[Andrew M.]

I will not use word aggressive but prompt, prompt reaction.
If you work with the camera and you see the processed material every day, and then you see good test but done in rush, you just want to comment.
I spent last week scaling, sharpening, CA correcting, testing different methods of processing RAW and I see all the mistakes I was making two weeks ago.

So I want to share with others so my time is not wasted just for me and I expect the same. The best example is hiss on release 15 from the mikes. I am about to play with audio and today’s posts about audio will save me a lot of frustration.

 

[Steve Sherrick]

Have you notices that most of the follow up posts to Adam's article follow the same theme, and have similar "generic" email addresses, that just happen to bounce?

I'm as passionate about education as much as I am about Red. That's why my posts are based upon scientific fact, and I try to teach when I write. You'll notice that you don't see that kind of posting from these trolls.

Graeme

You always back things up with data Graeme, and I like that. Because then we can take that and use it as a basis for our own testing.

I learned early on with forums, it's best to use your real name and then when you get in a real room with people it's the same kind of interaction. You are who you are, whether on a forum or face to face with people.

 

[Graeme Nattress]

And through trade shows, you get to meet us, and talk with us face to face. I'm sure a lot of people on this forum have either met me, or spoke to me on the phone, or had extensive email conversations. That's the real difference.

And I don't take bullshit from anyone. And I won't tell lies to sell cameras either. Right from very first being asked about bayer patterns, back on DVXUser, I said >70% linear resolution factor, but cannot be as high as 100% due to optical low pass filtering. And today, even with our Redcode compression, and optical low pass filtering, we're still >70% and I can just see detail out to 78%. A guy on CML measured that same 3.2k as well, so it's not just me either Chroma resolution tops out at 2k, but good algorithms a) reduce any chroma moire to practical invisibility and b) make the chroma channels look really nice and totally co-sited with the luma.

In the end, you get a great image, that stands up to being blown up onto a 60ft screen. I remember at IBC we were getting ready for our 4k presentation, and on the same screen was some HD footage before us. There was no comparison - it looked like, well, big VHS.

Graeme

 

[Deanan]

We appreciate it alot when everyone shares their experiences and successful or unsuccessful workflows. There are many ways to do things wrong and lots of ways to do things right too. Every pipeline has different goals, biases and priorities. What one person thinks is the right way might now work for someone else. We see this all the time with post vs vfx vs indie and so on.

 

[Uli Plank]

In the end, you get a great image, that stands up to being blown up onto a 60ft screen. I remember at IBC we were getting ready for our 4k presentation, and on the same screen was some HD footage before us. There was no comparison - it looked like, well, big VHS.e

I've watched through both screenings at IBC, the other one were the Japanese with their "Ultra HD" at 8K – And Graeme is absolutely right, you might have been able to count the hairs on the pedestrians heads in a wide shot from a high-rise, but the image looked like cheap VHS! Oversharpening, burnt-out highlights and plastic colors. The footage from "Mysterium" right next to it looked like some of the best shots from my valued Canon SLR – but these were moving images! The next day I started bugging my art universities administration to order a RED…

Image quality is much more than pixel counts. According to serious tests, 35mm at the best cinemas is barely resolving 1000 lines, but did you enjoy movies in your life? Sure you did!

Regards,

Uli

 

[Steve Sherrick]

And through trade shows, you get to meet us, and talk with us face to face. I'm sure a lot of people on this forum have either met me, or spoke to me on the phone, or had extensive email conversations. That's the real difference.

And I don't take bullshit from anyone. And I won't tell lies to sell cameras either. Right from very first being asked about bayer patterns, back on DVXUser, I said >70% linear resolution factor, but cannot be as high as 100% due to optical low pass filtering. And today, even with our Redcode compression, and optical low pass filtering, we're still >70% and I can just see detail out to 78%. A guy on CML measured that same 3.2k as well, so it's not just me either Chroma resolution tops out at 2k, but good algorithms a) reduce any chroma moire to practical invisibility and b) make the chroma channels look really nice and totally co-sited with the luma.

In the end, you get a great image, that stands up to being blown up onto a 60ft screen. I remember at IBC we were getting ready for our 4k presentation, and on the same screen was some HD footage before us. There was no comparison - it looked like, well, big VHS.

Graeme

I guess that is why I get a bit frustrated when people start getting personal and questioning your credibility and accuse you of toting the company line. Because you have always been incredibly helpful and honest dating back to when I was purchasing your plugins for FCP.

For someone to say that the camera is worthless and should never be used on a film shoot seemed really bizarre. But you may be onto something about the real identities of these people.

Anyway, keep up the good work and look forward to the upcoming firmware.

 

[D Namir]

For anyone interested in putting Leica lenses on their RED camera, Abel has a set of rehoused Leicas in PL mount that are excellent glass.

For anyone in Israel interested in putting Leica lenses on their Red camera or on any other motion picture cameras or Phantom, there is a set of rehoused Leicas in PL mount which include a 15m 19mm 21mm 24mm 28mm 35mm 50mm 75mm 100mm (macro) and 135mm (macro)

 

[Tom Lowe]

So 2.5k, that sounds about right.

 

[Graeme Nattress]

2.5k is about right for an unfocused image or poorer lens. I can see about 3.2k on the tests I've had Jarred and Jim shoot for me. I can see aliasing out to 4k, but only at a very, very reduced level, and I don't see any aliasing at 2k resolution, so I think we've got that balance very good.

Graeme