Clarification, part 2 - HD-SDI

[Gunleik Groven]

What I thought was that one could use the dual HD-SDI outputs for recording to f. ex. an HDCAM SR solution @ fullraster 1080 4:4:4

This would be an excellent argument for those who are highly sceptical to wavelet compression (and I know a few) but would hire an F23 to do the same thing, as you'd get true s35mm carracterristics and full raster with Red.

Also in our studio, we have set up 5 XSERVES with SXSAN and XRAID, where 3 of the xserves basically are recorders for uncompressed HD through HD-SDI.

I can easily see scenarios where we'd like to keep this workflow.

The discussion here is not whether 1080 is better than 4k, it isn't.
I don't question the compression scheeme at all

But I need arguments towards customers who'd otherwise go F23 and it would be an advantage NOT to scrap a working workflow when "the old way" just is more practical.

But if the HD-SDI quality is not full, these thoughts would cause some pain...

Just to be clear:
I have bought into the Red workflow arguments (as far as I have understood them...), but I guess for a while I'll have to deal with both workflows and people who haven't...

Gunleik

 

[Fredrik Callinggard]

I understand Gunleik but in reality this should be an easy discussion. Just explain to the client that shooting to HDCam essentially makes it a video format. While shooting RAW makes it similar to film negative and therefor is more flexible in post production. The latitude of RAW is closer to film, that's why you want to shoot it in RAW and preferably 4K, so you can downres. The client would have a greater choice afterwards in dealing with the picture.

fred

 

[Kjetil Haugen]

Hi gunleik

I was with some DP's in Lillehammer yesterday testing the F23 against the Silicon Imaging SI2K. Drove through Espa by the way... (måtte jo kjøpe 5 verdens beste boller og kaffe for 50kr på benser'n.)

We shot outside and in studio with the cameras side by side and color graded in Speedgrade later with the images side-by-side on a calibrated projector.

In this test I'd say the F23 was a sure winner with beautiful colors and a natural feel. That's not to say that the SI2K didn't perform, but given the nature of cmos sensors we had a hard time getting color into the image (to match the F23) as it was more flat from the start. We had a great colorist and DIT working the footage, and they managed to get close, but had to go into secondaries and tweek awhile. Skintones was difficult. Their thoughts and comments were very educational.

So I don't think it would be the wavelet codec (SI2K used cineform) that would make people "go F23". More likely it would be because of F23's 3ccd's and the color management straight out of the sensor.
That's my opinion anyway.

We also watched film-outs of viper and SI2K tests in a 35mm projection theater at the filmschool - also compared these test to other 35mm.
I'd say they all looked like film to me. I wouldn't had been able to tell the difference had they not told me what was what. Very encouraging!

I'm so looking forward to doing similar tests with Red when I get mine. Even though SI2K is very similar to Red (cmos, cineform, PL-mount etc) I can't say I'm worried about saturation issues judging from the images I've seen on this forum and the frame grabes I've played with. Also I'd think Red would have a huge advantage in the noise department, with it's 4k resoultion.

 

[Gunleik Groven]

Agreed, but some will rather stick to a tried workflow, and HD-SDIs should ease that argument, and open the door for:

But hey, we can do it even better!

It's gonna cost you some more in transcoding times atnd CC, compared to the workflow which is up and running, but you'll just gain so much more in image quality. And if you still choose to go the "old route" (dual HDSDI to a raid) you'll still get the advantage of an downsampled 4k image and full 35mm sensor with the 35mm DOF carracterristics.

It would be a total win-win.

Edit after reading Kjetils post...
Var invitert opp, men hadde ikke mulighet...

Very interesting. I'm so curious about side buy side comparison of Red and F23!


Gunleik

 

[Laco Zamba]

Hi Kjetil, which lenses have been used?

 

[Kjetil Haugen]

Hi Kjetil, which lenses have been used?

Cooke S4 primes for the SI2K and Zeiss Digiprimes for the F23.

Watching the filmouts of the SI2K they had one shot in there with a nikon lens and that shot was the only one I noticed as having a less "film-quality" look...

 

[Jason Rodriguez]

In this test I'd say the F23 was a sure winner with beautiful colors and a natural feel. That's not to say that the SI2K didn't perform, but given the nature of cmos sensors we had a hard time getting color into the image (to match the F23) as it was more flat from the start. We had a great colorist and DIT working the footage, and they managed to get close, but had to go into secondaries and tweek awhile. Skintones was difficult. Their thoughts and comments were very educational

Just curious, what did you grade from the SI-2K? Did you grade the pure "RAW" data with all the color metadata turned off? If so, it will be nearly impossible to get the correct colors without secondaries, and specific color manipulation because you must have a matrix applied in order to get the color vectors aligned properly. Of course that being said, the color-separation from a 3-chip camera will always be better than a single-sensor camera that naturally has some color cross-talk from the bayer pattern.

 

[Stephen Williams]

I'm so curious about side buy side comparison of Red and F23!


Gunleik

Hi Gunleik,

I think when #65 is delivered you will see such comparisons.

Stephen

 

[Kjetil Haugen]

Just curious, what did you grade from the SI-2K? Did you grade the pure "RAW" data with all the color metadata turned off? If so, it will be nearly impossible to get the correct colors without secondaries, and specific color manipulation because you must have a matrix applied in order to get the color vectors aligned properly. Of course that being said, the color-separation from a 3-chip camera will always be better than a single-sensor camera that naturally has some color cross-talk from the bayer pattern.

The raw cineform files were rendered to 10bit dpx before color grading. I don't think they had software like redalert or redcine adjusting the raw material before converting to dpx. I'm unsure, I was in the theater watching film-outs while they prepared the files. They used AE for dpx convertion, though.

Nevertheless, the F23 and SI2K are both great cameras. Something really cool with the SI2K was that you could take out the front of the camera and you'll have a small box to attach lenses to that you really CAN put in your pocket. Really great for car interiors etc. Like the DIT said. We were comparing apples and oranges. If we add Red to the mix, we'd be comparing apples, oranges and bananas... They are all different cameras.

I'm even more exited now to see what Red can do. Watching these cameras, I'm not worried about 35mm being superior to digital anymore.

 

[TimPipher]

What I thought was that one could use the dual HD-SDI outputs for recording to f. ex. an HDCAM SR solution @ fullraster 1080 4:4:4

This would be an excellent argument for those who are highly sceptical to wavelet compression (and I know a few) but would hire an F23 to do the same thing, as you'd get true s35mm carracterristics and full raster with Red.

Also in our studio, we have set up 5 XSERVES with SXSAN and XRAID, where 3 of the xserves basically are recorders for uncompressed HD through HD-SDI.

I can easily see scenarios where we'd like to keep this workflow.

The discussion here is not whether 1080 is better than 4k, it isn't.
I don't question the compression scheeme at all

But I need arguments towards customers who'd otherwise go F23 and it would be an advantage NOT to scrap a working workflow when "the old way" just is more practical.

But if the HD-SDI quality is not full, these thoughts would cause some pain...

Just to be clear:
I have bought into the Red workflow arguments (as far as I have understood them...), but I guess for a while I'll have to deal with both workflows and people who haven't...

Gunleik

Hello Gunleik. This is the workflow I want, too, using three Reds with a swticher in a studio configuration using an HDCAM SR deck or Wafian 2 recorder. This is possible, correct? Or have you heard it now can't be done with Red?

Also, in this configuration using the Red 18-50 mm zoom, would you still get the true s35 mm characteristics including DOF or would other lenses be needed?

Thanks,

Tim

 

[Gunleik Groven]

As to the DOF: You're fine

As to the output from dual HD-SDI, the quality of that output is maybe not as clear, as far as I have understood.

I don't know if
a) I've misunderstood
b) This is a temporal problem (most likely is, if not a)
c) This is a feature that just has fallen off the map for the time being. Guess not, though, why else would one have dual HD-SDI capacity?

Anyway, I'm curious about the current roadmap re HD-SDI output from camera.

Gunleik

 

[RivaiC]

Been thinking of this for quite sometimes, good question. Some need to be clear before i'm go to red.

We're mostly do a long form project. It would be damn hard and long to wait for transcoding. The only way i can think of is to record on HDCAM SR or D5. But again not too sure. I hope some RED people here can give some insight.

 

[Gunleik Groven]

As to image quality, I'm quite sure the 4k -> downsample in post to whatever is the best.

As to time spent on a project...

Quite a few projects would become profitabele from recording 1080 through HD_SDI, that wouldn't really make me any cash in 4k...

Or I'd just have to pass'em on.

Gunleik