PDA

View Full Version : Some Decembers shows and some things learned.



I Bloom
12-21-2007, 01:08 PM
In the past two weeks and a half I was able to complete principle photography on three small productions with Red #87 and Vladimir Eugene. I want to post some stills and some quicktimes but I need to do that with the producers and directors I've been working with so that their rights are respected.

I'm posting here in a similar way to my first experience using Red. But I want to note that all of the issues I encountered are known or have already been addressed by Red. As you know Red has made many changes and improvements to the cameras they are about to ship that are not reflected in the body that we are using. Likewise we chose to keep build 10 (v 1.6.4) on our camera, Red had already released build 11.

Here is the upshot. Compared to my first experience shooting on build
6. Build 10 is essentially a different camera, better stronger, faster with more features that made accurately focusing and exposing possible.

Build 10 is like being captain hook and finding out you can get the hand replacement from Star Wars.

But Build 10 is also like being Luke Skywalker, and deciding "I'm gonna roll with the hook."

We had no camera related issues during our first five days of shooting and over the course of 12 days on set we had no issues that I could trace back to firmware (with two non-critical exceptions). I shot in all kinds of conditions, especially some wet and snowy night and day exteriors where the temperature was sometimes hovering around and below the freezing point.

The purpose of this thread is to share my experiences, my problems and the solutions I found. I'm trying to keep things as objective as possible, this is what happened and this is how to avoid these pitfalls.

Production Title: Glossy Veneer, Short Film
Director: Tom Payton
Producers: Alex Cripe, Andrew Renzi
DP: Ibloom
DIT: Vladimir Eugene
Gaffer: Meg Schrock (meg at megschrock dt cm, http://www.megschrock.com)
Best Boy Electric: Adam Benlifer
Key Grip: John Shim (also can do in-key-grip-playback of important takes.)
Best Boy Grip: Travis Tips
Rigging Grip: Luis Colone
1st AC: Alex Peterson (Awesome, yet quietly putting fear into the hearts of production)
Lenses: Zeiss Superspeed MK3's
Camera Support and Lenses: Du-All Camera http://www.duallcamera.com (he's got brand new S4's now, but scheduling didn't work for us.)
Delivery Format: Film-out.

Our first project was a five day short film entitled "Glossy Veneer". The film included among other actors, Lucy Devito, daughter of Danny and Rhea. Shot on location in Greenwhich Connecticut with a mixture of interior and exterior shots.

The first problem I ran into was that the flange focal depth was out on our camera, and I couldn't get it reset because without a depth gauge the process of shimming the camera is aparently very time consuming, essentially trial and error. As you know, Red is replacing the aluminum mount that we were using with a user adjustable stainless steel mount and replacing their first 100 cameras, in part to address this potential problem. The solution to this problem is fairly simple but not desirable: The focal depth is set to something, just not to spec, so make a new ring on each lense with camera tape and lay down new marks. Took me about three hours. Not fun but it definitely works. I was lucky because I took the same package on three productions, so I only had to do this once.

While I was calibrating the lenses I was able to use Graeme's magic focus assist in combination with image zoom. Pointing it at a focus chart, I have mixed feelings. I feel first and formost that I don't understand how it works. So here are some perplexing things:

-If you focus up on a chart and make it sharp, you will see very high peaks. However if you then dim the lights, the peaks will decrease as well.

-If you focus up on something that out-resolves the sensor it will read as soft. This is easy to tell on a focus chart, harder in practical shooting.

In practical shooting this makes the meter difficult to understand. As I understand it a tall peak definitely means something is in focus (though I aim to test this further). But a low peak does not mean that something is out of focus. Focus is relative to contrast here. So unfortunately: perhaps not false positives but many false negatives.

I found that in normally trying to use the assist to make sure that the focus was dead on the eyes. On a closeup we would move the assist bar to the position of the eyes, but it wouldn't pick them up even when I knew for certain they were sharp.

I found that the red bar is terrible for operating since it messes deeply with your sense of composition. Best to leave it off.

I would love an alternative to the joystick. In the meantime I think all menu's should roll over, so that if the joystick isn't responding in one direction, you can just roll over in the other direction.
I still occasionally had difficulties getting the Red LCD up, but the situation was greatly improved. A quick solution to this problem is to jump into the menu and switch to EVF and switch back. That does the trick.

We had another problem, which I determined was not caused by the camera. I'm including it here as a cautionary tale:

Our abbey singer shot for the whole show was an 80 ft dolly leveled to a 5ft elevation change, with limited time at the location, a steel welding facility in Connecticut. This was likely the most expensive shot we did. John and Travis did an incredible job of making this happen. While Meg and crew overcame a herculean task of lighting the huge background in the rain at night. After we completed about 8 takes with the dolly and verified them with director via in camera playback we passed off the card and hurried to move on break down the dolly and relight for a wide establishing pan of a van driving into the location.

After we completed the final shot I came back to the DIT area, and discovered we had a situation. We had unfortunately pulled a CF card out of the reader while it was in the process of mounting. The result of this was that many of the files on the disk including all of the R3D files appeared as Zero KB in Finder. I felt at that time as if I had just drank about 10 shots of espresso. I immediately went into the terminal and looked into the Volume, same story. It appeared our FAT32 file allocation table had been damaged. I put the card into the camera, and it read it as full, but could not play back the clips. Unfortunately reshooting was not an option at that time.

I want to stress that this was not a camera related issue, as far as I can tell the camera functioned properly, created proper files and was able to play back clips several times before we attempted the transfer. This was a transfer issue that taught me that CF cards are not as robust as I had thought. The lesson here is never dismount a card manually, if you attempt to mount a card and it does not show up. Shut down the system, remove the card and reboot. And never, never, never move on without verifying that important takes have properly transferred and will play back appropriately from your archive drives. Transfer is HIGH RISK. You can't blame on the DIT, it is the production's job to verify that the transfer has occured before the move on from an expensive setup. You need to adopt a set protocol that checks the transfer, the same way we check the gate on film. (By the way, Eric from Red was extremely helpful to us, I never got a chance to get back to you, thanks.)

I would suggest also using a Lexar Firewire 800 reader, we were at the time using a USB Multi port reader, for various reasons. I have seen the Lexar mount faster and my gut says that the Lexar is more robust.

If Red could create a CF solution that had a locking mechanism like P2 it would be worth the extra expense.

After looking into several file recovery software options we eventually decided to take the card to professionals. After getting an evaluation from several companies we settled on sending the card to a company in Florida that specializes in CF cards, http://www.lc-tech.com/ I am awaiting to here back to them if the print take can be recovered. It's my fault, I was riding the dolly thinking, "This one's going on my reel."

I Bloom
12-21-2007, 01:09 PM
Production Title: Unnamed
Producer Directors, Crew: Unnamed for political reasons regarding ties with celluloid manufacturer.
Delivery Format: Film Out
I'm leaving out the details here because the originating format may cause problems for the production until after the release. In a few months I can explain more. Suffice it to say I shot for two days on a sound stage in front of a rear projection background.

I came up against two obstacles on this shoot, the first is small and straightforward but when unexpected can hurt pretty hard. The battery started occasionally loosening on the V-mount. Causing the camera to trip out and the rear LCD to go white. This also seems to happen on the Red charger, such that you think its on and charging and it's not. Solution: Pop the battery off and on again and reboot. Old news I hear.

Second issue. We had originally considered using greenscreen but because of foreground transparency and refraction the VFX guys said we should do rear projection practically. Issue: I don't think there is a rear projection system we could obtain that is ready for Red 24fps. I tested both an LCD projector, and a DLP projector. Unfortunately the LCD image was immediately revealed to look pixelated, yet had no flicker issues. The DLP projector looked great in terms of quality but had a flicker issue, DLP is constantly cycling through RGB images each exposing for a little less than a third of a frame interval. In order for this not to be a problem the projector would need to display all three colors between the readout periods of the sensor. I believe the model we were using was running 29.97, so we had an issue. I know its a shutter issue because you can put the screen way out of focus and you will still see hard vertical cuts exposing either red green or blue.

After testing both types of projection, the producers decided they were going to use DLP regardless of the flicker, which wasn't terrible. But it forced me to shoot the whole show at 1/24th second. Which is an f-ing terrible creative choice to be forced into. It's just how the cards came down, until a few days before we shot the plan was greenscreen, then suddenly we had to change, and there was no time to test as well as we should have, both options had issues. I can verify that DLP will not create the same issue on an HVX.

Aside from those things that the Red performed marvelously. Having a MacPro on the set with a 23 inch monitor is glorious with RedCine.

Production Title: Mates of State "Get Better"
Director: Daniel Garcia http://www.danielgarcia.tv
Producer: Mike Harry http://www.concentrate.tv
Executive Producer, Post Production: Steve Choo of My Active Driveway http://www.myactivedriveway.com
DP: ibloom
Steadicam Operator: David Ellis
Gaffer: Meg Schrock, Best Boy Electric: Adam Benlifer
Key Grip: Luis Colone, Best Boy Grip: Pedro Diaz
Grip: Jonathan Kohnken
Delivery Format: HDCAM

We also shot for two days for a the band called Mates of State, for there new album that should be released this spring. It was a wonderful production and my first time shooting for a band I had previously paid to go see. The band already leaked this blog entry:
http://babble.com/CS/blogs/bandonthediaperrun/

On set the camera performed wonderfully. I can't break the firmware. But on our second day we ran into a problem because of the cold. My fault for not reading carefully enough on this site: in preparation for the cold we had changed out the filters and set the camera into silent mode. We then proceeded to shoot outside for an entire day with absolutely no camera problems. We even accomplished a hood mount and a hostess tray shot with no issues. But as the shadows got long I think the temperature started to drop and we made the mistake of shooting a hard mounted steadycam shot off the back of a pickup without further prep work. I think the lowering temperatures and the windchill combined to push the camera over the edge. The result was codec blips, frames where part of the image doesn't seem to include all of the compression information. Jim had already warned us to put a sweater on it, so I can't blame the camera. However, my one criticsm of the camera from this perspective is that it seems too fault tolerant. I'm glad that it keeps rolling through these problems but I wish that it warned the operator. Once we became aware of the problem it was too late to reshoot. It is possible we can fix the blips in post, since this is a heavy VFX project but for various reasons we are now looking for a camera we can use to pickup these takes.

The post for this project is being conducted by My Active Driveway. A VFX post house in Soho. I met with the executive producer and producer from the company and took the opportunity to discuss with them what I've learned about working with Redcode. It seems clear to us that there are a lot of pathways to get to where they want to go but their situation opened my eyes to some of the issues that you are facing, if you shoot on Red and you want massive visual effects and heavy color work: without native Redcode support the pipeline is crippled. Scratch would be a great asset in these situations but scratch is not a full featured compositing system. Their main option if they want to maintain bit depth, is Tiff sequences or Cineform. I have a lot more to say about post, in the coming weeks, I'm continuing to learn a tremendous amount.
--------
One other thought: If you read this site closely, you will know that our first ever attempt to shoot on a Red about two months ago was derailed by a problem with the locking collar on our PL mount. That film was instead shot on HD. Later careful consideration brought us to an important conclusion. It was likely that during transport, our mount had been torced too far. I'm going to make a suggestion that you consider removing the fins from your locking collar. They look cool but it seems to me that there should be no situation where one needs to apply that much pressure to a mount. Instead having them off could prevent damage. Kind of like removing the handle from a tripod when you pack it and keeping the head unlocked. I'm interested in your thoughts on this.
---------
One other thing. Last week we also shot an interview with Keith Richards with three cameras including one Red. The audio component of the interview will be on Sirius Satellite radio. Keith's management is extremely tech savvy, and they knew all about the Red. I'm doing my best to clear a frame and post it here. Upshot: we spent two hours in a small room with Keith Richards telling stories. He is the coolest human I will ever meet.

I think we get the prize for first to shoot a rock legend on Red.

Thanks again to Red for this awesome camera.

IBloom

Jim Exton
12-21-2007, 01:30 PM
Thanks for the report. These kind of on location reports are extremely valuable to all of us.

LEON
12-21-2007, 01:30 PM
thank you so much for these hard lessons !
i"ll try to remember all of it, I am sure I would have fallen
in the card issue trap !

Patrick Tresch
12-21-2007, 01:53 PM
But as the shadows got long I think the temperature started to drop...

I love this sentence. Thanks for your report and poetry.

Pat:biggrin:

Cail Young
12-21-2007, 03:02 PM
and refraction the VFX guys said we should do rear projection practically. Issue: I don't think there is a rear projection system we could obtain that is ready for Red 24fps.

Either a higher-resolution LCD system (with a smaller pixel-pitch, i.e. 2K and above) or one of the newer DLPs that use a triple lighting system rather than a filter wheel which allows much faster cycling may be worth a look next time this comes up...

Seth Larney
12-22-2007, 04:32 AM
The post for this project is being conducted by My Active Driveway. A VFX post house in Soho. I met with the executive producer and producer from the company and took the opportunity to discuss with them what I've learned about working with Redcode. It seems clear to us that there are a lot of pathways to get to where they want to go but their situation opened my eyes to some of the issues that you are facing, if you shoot on Red and you want massive visual effects and heavy color work: without native Redcode support the pipeline is crippled. Scratch would be a great asset in these situations but scratch is not a full featured compositing system. Their main option if they want to maintain bit depth, is Tiff sequences or Cineform. I have a lot more to say about post, in the coming weeks, I'm continuing to learn a tremendous amount.

Thanks so much for your detailed report, iBloom. Lots of very good info here !

One note though. We are a Production company slash VFX/Post facility who does alot of 3D and 2D work and we would NEVER perform any VFX whatsoever without an uncompressed image sequence pipeline. No matter what the originating footage (HDV, DV, HDCAM, Film etc), our first step is always to convert into uncompressed DPX. I'm not sure why (with the exception of people doing this at home on desktops) anyone serious about VFX would do otherwise.

Now, I know what the datarate of RGB 4K is :) but if you want the best quality physicaly possible, you need to do things properly. I don't see the as being a problem at all.

Note : this is subjective, I know people will argue disk storage etc is more important, this is just how we operate.

Edit: I should add, this is how most pro VFX houses operate as well. I can understand why if you were wanting to composite at home etc, this might be a problem.

2nd Edit: I would also recommend trying one of the newer DLP projectors without the filter wheel if you have the need to shoot rear projection again.

Cheers,
Seth.

I Bloom
12-22-2007, 07:02 AM
Thanks so much for your detailed report, iBloom. Lots of very good info here !

One note though. We are a Production company slash VFX/Post facility who does alot of 3D and 2D work and we would NEVER perform any VFX whatsoever without an uncompressed image sequence pipeline. No matter what the originating footage (HDV, DV, HDCAM, Film etc), our first step is always to convert into uncompressed DPX. I'm not sure why (with the exception of people doing this at home on desktops) anyone serious about VFX would do otherwise.

Now, I know what the datarate of RGB 4K is :) but if you want the best quality physicaly possible, you need to do things properly. I don't see the as being a problem at all.

Note : this is subjective, I know people will argue disk storage etc is more important, this is just how we operate.

Edit: I should add, this is how most pro VFX houses operate as well. I can understand why if you were wanting to composite at home etc, this might be a problem.

2nd Edit: I would also recommend trying one of the newer DLP projectors without the filter wheel if you have the need to shoot rear projection again.

Cheers,
Seth.

Both good points, I think a newer DLP projector was out of reach.
As far as uncompressed goes, I think you are correct, but they aren't used to first creating a pull list and then downconverting. If you aren't constantly working with scanned film then that is what is new.

IBloom

Seth Larney
12-22-2007, 07:17 AM
Both good points, I think a newer DLP projector was out of reach.
As far as uncompressed goes, I think you are correct, but they aren't used to first creating a pull list and then downconverting. If you aren't constantly working with scanned film then that is what is new.

IBloom

Gotcha mate,

Seth.

PaulClements
12-22-2007, 08:20 AM
Thanks for all the info Ian. Useful reports as usual.
Cheers
Paul

Gavin Greenwalt
12-22-2007, 10:49 AM
Thanks so much for your detailed report, iBloom. Lots of very good info here !

One note though. We are a Production company slash VFX/Post facility who does alot of 3D and 2D work and we would NEVER perform any VFX whatsoever without an uncompressed image sequence pipeline.

Cheers,
Seth.

We do the same. But I've done Cineform tests and I would have no problem at all with a Cineform intermediate. The only thing that prevented us from acting on those tests was the cost. $1000 per seat can buy a lot of RAID.

I Bloom
12-22-2007, 02:50 PM
We do the same. But I've done Cineform tests and I would have no problem at all with a Cineform intermediate. The only thing that prevented us from acting on those tests was the cost. $1000 per seat can buy a lot of RAID.

It seems like Redcode-RGB would be a good alternative, whether made in camera or not.

IBloom

Seth Larney
12-23-2007, 04:11 AM
We do the same. But I've done Cineform tests and I would have no problem at all with a Cineform intermediate. The only thing that prevented us from acting on those tests was the cost. $1000 per seat can buy a lot of RAID.

Cineform used to actually be an essential part of our pipeline, just not our VFX pipeline.

We would online with cineform and then render out to uncompressed dpx for the VFX part of the pipeline.

Cineform is an absolutely MAGIC codec (the only codec so far that I would ever use as an intermediate). We just found that the decoding power required once you start layering it down in a composite was not worth it. Hence, Cineform for the online, DPX for VFX.

Now we do DPX the whole way.

Edit : For anyone looking for a high quality intermediate and simply can't afford the speed/space of uncompressed, I would absolutely recommed Cineform !

Cheers,
Seth.

I Bloom
12-29-2007, 01:42 PM
Great success!

We were able to recover all of our lost clips.
The guys at LC Tech in Florida got the job done. http://www.lctech.com.
Our contact there was Ray Hasinger. Ray let me know that they are going to include .R3D support in the next build of their photo recovery software.

Here's a Tiff from the recovered files:
http://www.redhax.net/007_01305.tiff
And a quicktime from one of the shots we lost.
http://www.redhax.net/GlossyVeneerH264.mov
Please do mirror.

Enjoy.
Ian

C.H.Haskell
12-29-2007, 02:24 PM
As usual thanks for sharing Bloom. :)