Derek Schweickart

Well, the simplest and most meaningful way to think about filmstream is to consider that a CCD can only do one thing, it reacts to light. Any gamma, gain or color correction applied does not change how the chip reacts. Why correct the image in camera rather than acquire the most raw data of the light hitting the sensor, and correct it later? I believe the DP would want to acquire the most information about the light in a scene as possible. If you are very interested in painting the chips on set, then Filmstream is not for you. As far as the log singal is concerned, Thomson decided it is a better way to use the 12 bit data since they are forced to scale it down to 10 bits. I don't have a definitive answer on the highlight question yet, but yes the white point is the same log or no log, and no, highlight detail does not exactly equate dynamic range. Anyway, Filmstream allows the DP to shoot as if he were shooting a particular film stock. He knows how it reacts to light and he works with it. It makes for a very fast moving schedule because there is no painting on set, which of course does not make DITs happy, I understand the resistance. It's a good camera, it's light and easy to use, and produces a great picture, and with filmstream you can point, focus and shoot and be sure you're getting as good a picture out of the camera as possible. That's a bit simplisitc but there's the idea.

Mike, I appreciate the challenge, this is what the forum is about. Unfortunately, I have been out of town lately and haven't been able to address your questions. I intend to do some more reseach to give you a more complete answer, but for the moment I can say that "good gammas" can achieve much of the same effect as the log output but the argument for using log is that it is a constant value and gives you good results without painting the image, and tweaking the gamma settings for each shot. This is a workflow option and is debatable. There is also an argument in favor of shooting log when outputting to film, because you can reliably apply a reverse log function to the signal and recreate an accurate 12 bit linear signal. Let's continue this debate, I'll see if I can't find some more concrete answers. Derek

Well because the color space is still 4:4:4 RGB, the same as the Sony F950. Filmstream is a mode where all the color processing circuits, matrixes and gamma are turned off and at the same time the signal undergoes a 12 bit to 10 bit logarithmic conversion instead of the normal linear scaling. This results in more exposure latitude and a less noisy image. It also gives you more options as far as what you look you want in post because the image starts out flat and low contrast but has enough bit depth to be able to push it around any which way you may want to go. In any case, Filmstream really shines when trying to hang on to highlights or pull detail out of the lower tones. If you have any more questions, feel free to ask. Derek

I want to address a couple of the topics posted here. One, yes Thomson is making a concerted effort to distance the Viper from the term "Video." I don't think that this is as hypocritical a stance as some are making it out to be. The truth is that, of course the Viper is not film, but in "filmstream" it bypasses most of the tools that are essential to video cameras. Namely, there is no gamma curves or color processing applied to the image. I personally think it is better to think of the Viper in terms of being part of a digital intermediate process than being a video camera. At the Camera House we use the term "Digital" rather than Video to differentiate the Viper workflow from typical HD. Second, the rental price of the Viper is less than an Arricam studio and comprable to a 435 package. The support/recording/viewing equipment rentals can add up but this must be seen on a case by case basis, and when using the S-Two recording system with a proper opertator you replace the need for video assist and loading. It is hard to come up with hard numbers to compare film vs. digital but from what I can gather the price of the packages and personnel required are about even before you consider the savings in film stock, developing and perhaps digital scanning. Now, the long format workflow when using S-Two devices, consists of backing up recorded material to LT02 (soon LT03) data tape. This is an acceptable deliverable and is in effect your digital negative. You can cut from lower res dailies then conform back to your original material. Again I don't know what exactly the numbers are for post-production but on a long format project I believe there are great savings. Silence becomes you was recorded onto S-Two and archived using S-Two's A-dock. They were very pleased and if they say it ended up being cheaper, I don't doubt it. I hope someone found this helpful. Derek

Thanks for the info. I'm wondering what you think of the HDlink if you've used it before. I've used Ecinema and AJA DVI converters for viewing but not the Blackmagic.

As far as I know, the Sony devices do not. As far as HDlink goes, I have not used that card and do not know that much about it. I can't imagine it would be easy to put LUTs in a hardware device because they need to be flexible and customizable. It's much more of a software concern.

We currently own three. We own two S-Two DFR's for capture. I am not in charge of purchasing so I am not sure what is in the pipeline, but that's what we have right now.

Sounds good. We'll be here

Any time you want to come by the Camera House I would be happy to give you a demo. Also I'll try my best to answer any questions you want to post to this site.

Yes, the Viper can record to the SRW-1 or SRW-5000 in filmstream mode. It is still 10bit information so even though the signal undergoes a logarithmic conversion you're still dealing with the same bit depth. S2 has a tape backup system called an A-Dock combined with a Scalar LTO2 (soon to be LTO3) data tape jukebox. The LT0 format is made by Fuji and is an approved deliverable format. It is widely used for data backup by banks and insurance companies and many other institutions dealing with data archiving. The good thing about data tape is that it is cheap and lightweight, like video tape, but is still uncompressed. Also, LTO readers are cheap. I think HP makes one for a couple grand. When shooting in 2:37 mode, the Viper does not output a squeezed image on the viewing port, but it does squeeze the viewfinder image. The S-Two DFR has a built in squeeze to output a corrected image for viewing on all of its outputs. The only real problem is if the operator wants to use a monitor instead of the eyepiece. It would require a video feed back from the DFR. Realistically though, if this is the case, a second cable could be taped (or snaked) with the mulitcore cable from the camera since it has to be there anyway. Hope this helped.

When dealing with HD, people seem to focus so much on what that camera is doing that they forget that the recording technology is just as important. S-two makes a great recorder and the guys there are working very hard to create a usable digital workflow for high end production. I'll be happy to answer any questions about the S-Two system, but I admit to not knowing all the other competing systems as well. What else is out there that can record uncompressed 4:4:4 RBG?

I should be there in the S-Two booth for a couple of days. Not sure which ones just yet.

Because things seem to be moving so fast in the world of digital I want to start a thread here to put news about the Viper system. For me this includes necessarily news about the S-Two recording device. As many of you know, Thomson has announced their onboard flash RAM recording device called the "Venom". Check out the official page for more info. http://www.thomsongrassvalley.com/products/cameras/venom/ For me this has been a large missing link in this stage of digital cinematography, so onboard recording of uncompressed Viper log images is a huge deal. S-Two has developped a button box for the DFR so that laptop control is no longer necessary. For those of you interested in more information about S-Two, I will do my best to respond in the "S-Two" thread. Another sidenote, the Camera House has made a deal of physical add-ons to the Viper including a power breakout box (24 volt Fischer 2-pin to 2 x 24V Fischer 3-pin and 2 x 12V Fischer 11-pin), onboard battery (24V), handheld base plate top, studio base plate top, and a series of modular quick release shoes to quickly organize and rearrange your accessories (Onboard monitor, Lens Control, Arri style carrying handle, etc.). These were all much needed modifications and they enable the camera to function much more readily as a hand-held camera. Combined with the Venom, you will find the Viper a much changed system come Summertime when Venom becomes available for use. Derek Schweickart Digital Tech The Camera House 7351 Fulton ave. North Hollywood CA 91605 818 997 3802

Elhanan, No Venom yet. Thomson will unveil the Venom at NAB and they should start selling them by summer. So they say. We plan on getting the Venom as soon as we can. You are absolutely right as far as the workflow goes. There would need to be two Venom per camera, one for recording and one for transfer. I think two is all that is necessary but that depends on how the Venom data is transferred. Dumping to the S-Two will be real time because you'll basically playback from the Venom onto the DFR. I don't know if there's a data transfer option on the Venom, but that could be longer than real time, and in the case of long steadicam shots, it may be necessary to have more than two Venoms. I hope people will use the Venom only when cabling is absolutely not an option because I am sure that the DFR affords you many more options.

The idea behind the Viper camera is flexibility. You can certainly use it as a normal High Def camera in its 4:2:2 Y Cr Cb mode (like the F900), or its 4:4:4 RBG mode (like the (F950). When in those modes it has full video control. The added advantage is being able to forfeit this control and record raw data in 4:4:4 10 bit log color space. Shooting 10 bit log essentially gives you and extra 2 stops in your highlights (two real stops, not like cranking up your knee) and I think it's about a stop and a half in your blacks. Like Claudio says, it's always better to give your blacks a little more light then crush them down later. In fact, when shooting raw the Viper will not produce a pure black, they tend to sit around 12% on a full range waveform monitor. This is because the black we are used to is a function of the gamma applied to the image. "what is the electronical control you have during the film sream mode? Even if it is raw signal 4.4.4 how does the camera react to high and low lights, what about the chroma? Whith the hd900 we control the picture with a 24''monitor and oscilovectorscope how do you control it on a viper if you get a sort of downconverted signal?" In filmstream you can modify the viewing channel which is a Y Cr Cb signal. Otherwise you can simply point, focus and expose for 320 ASA (400 in daylight). It is a good idea to monitor the signal with a 4:4:4 waveform such as the Omnitek, but you can pretty much rely on 320 ASA to be accurate. What I have found is that when properly exposing an image on the Viper for 320 ASA the waveform functions somewhat differently because you are dealing with a logarithmic signal which is not balanced. Also the green channel is always higher than the red and blue because there is no internal correction, and green light is always more prominent. What is interesting is that 18% grey falls at 25% on the scope. A white card falls at 50%. Above 50% you have 2-3 stops before you start to clip. Even though the raw image seems underexposed, when you balance out the colors and add gamma in post, the image looks correct and you end up with much greater latitude than if you shot in a non-logarithmic mode. Filmstream is not a WYSIWYG system, but once you understand what it is doing it is a very simple system to use. The best is to pull frames off of the S-two recorder and tweak them in Photoshop, or Final Touch HD. At the Camera House we are working on building an extensive set of Look up tables so you can get a good sense of what your scene can look like when shooting in Filmstream mode. These would be applicable on the fly and only affects the signal to the monitor. While I'm at it, any feedback on this subject would be much appreciated. Derek Schweickart Digital Tech The Camera House 7351 Fulton ave. North Hollywood, CA 91605-4114 818 997 3802