SI-2K and HDR?

[Alex Raskin]

I have not tried anything yet, but this (non-SI) info may be an inspiration:

ProVideo Coalition.com: Camera Log by Adam Wilt | Founder | Pro Cameras, HDV Camera, HD Camera, Sony, Panasonic, JVC, RED, Video Camera Reviews

Ask not what SI do for HDR, but what HDR can do for SI.

Or something like that.

Seriously, is there a way to increase dynamic range of the SI2K images?

[Bob Hart]

My imagining is that it might almost be already there with the 3D capability. An optical splitter for the SI2K mini pair might be an easier do than for other camera types.

We might someday down the track see a RED style "twice the framerate alternate frames thing" for individual camera heads evolve and come through as a major software update named DVR3.0. Otherwise HDR and 3D may not be able to co-exist except as in 4 camera heads with two optical splitters ????

For a small production run of software and hardware, the unit costs will be pretty high and will likely chase many people away.

As much as I like very much DVR2.0, I simply cannot afford it right now.

[Alex Raskin]

What's new/likeable in DVR 2.0? I was not following the developments since I'm pretty happy with the DVR soft that came with the cam a year ago.

[Bob Hart]

Alex.


I have only had a very short play with it so far. My impression is that DVR2.0 seems to be a lot smoother somehow. It also seems to hang onto higher quality settings in adaptive mode for longer, so the codec must now be more efficient.

The function tiles and screens are similar but there are some differences to improve convenience.

There is a cool feature where the start of your clip has a slate which you can program for name, scene and shot number. There is an automatic repair for munted clips like when some fool unplugs your power before the camera has time to tail the file. It commences after next boot-up.

I am not sure whether this is always on by default or whether it can be selected "off". In the middle of an intensive shoot, it might be a bit of a hassle waiting for the previous take to be repaired by the camera.

Give Ari at SI an email. He let me access a demo version which I downloaded and copied to a thumbdrive and installed over DVR1.

Just make sure you have a working thumbdrive image to restore your original system afterwards as the demo clips are watermarked.

The default filetype appears to be ".mov". I could not find ".avi" but it could have been renamed to another function.

[Alex Raskin]

I think I'll stick to my guns for now. If it ain't broken...

But thanks a lot for the detailed explanation of the differences.

BTW, I keep my quality settings fixed at Quality 3, which corresponds to FilmScan1 in Cineform's general codec. All is well.

[Mark Deeble]

Bob

Does DVR 2.0 give you a greater dynamic range? Shooting exteriors in Africa in uncontrolled conditions, I need all the DR I can get...
I guess a 2D HDR image could be achieved currently by using a pair of minis on a beamsplitter rig with an IA of zero, but that is pretty cumbersome and expensive - 2X camera & lens, rig etc.

[Alex Raskin]

I'd imagine that dynamic range is limited by SI2K's sensor, not by the codec or DRV...

[Bob Hart]

Alex and Mark.


I was about to say the same thing, my imagining is that the dynamic range is determined by hardware.

However there appears to be something to be gained by choosing the option of recording uncompressed. I understand there is about 12 stops of dynamic range however maybe the exposures have to be managed much more critically and the recording system and drives have to be capable of higher data flow.

In my early playings with the demo camera/recorder, I mismanaged it and was recording "I think" to uncompressed or something weird and ended up clipping the green channel by setting the exposure a bit too hot. I don't know what settings I used. I was experimenting with direct relay and a Letus Extreme groundglass adaptor at the time.

I would hope that there is enough headroom in the system to eventually allow a RED style alternating frames of different exposure settings. Apparently the RED can do something tricky like 48 frames per second with each second frame being the equivalent of a few stops high or low to yield 24 frames per second. I imagine the trade-off piper to be paid would be increased CMOS motion artifacts.

I guess the next new RED trick will be a mod on the bayer mask with twice the pixel count on the sensor and one set being masked to a few stops darker, reading out then separating into two frames.

This clip is one which Jason Rodriguez advised me on regarding green channel clipping.

SI2K - LETUS EXTREME TEST. By Bob Hart On ExposureRoom

For shooting in T16-T22 high intensity outdoors lighting conditions with the single camera, I would be tempted to investigate ND filters and the infra red filter.

Mark. Are you using Super16mm primes or a B4-mount HD zoom lens?

[Alex Raskin]

Originally Posted by Bob Hart

Apparently the RED can do something tricky like 48 frames per second with each second frame being the equivalent of a few stops high or low to yield 24 frames per second. I imagine the trade-off piper to be paid would be increased CMOS motion artifacts.

Haven't noticed any especially bad motion artifacts on RED footage - it is downloadable from the link at the top of this thread. Whatever they do, it works.

I'd really want to see two videos - one with, one without HDR. What they did in the article, is linked to a video with HDR; and to a still photo without HDR. Not a great comparison.

The reason I want two videos is simple. I downloaded RED HDRx video and graded it to my liking.

Now, what am I suppose to compare it to? It looks OK on its own. Would it look dramatically better than non-HDR one? I don't know - there's no reference source for non-HDR video in that article...

[Mark Deeble]

Bob

I've been using a variety of lenses - a Canon 11-165 mm PL mount S16 zoom, a Zeiss 12-120 T2 S16 zoom (Optex conversion), and Nikon 300 2.8 and 500 4.5. S16 primes have been Angenieux 5.9, Kern 10mm, 26mm macro etc. I've not used any B4 mount zooms with an Abacus adapter - have you (?) Do you get a noticeable drop in quality? I would like to use something like the Canon HJ28 which is a great lens for wildlife - but feel I need to test the adapter first.
As I see it - I need to compare like with like - ie the PL cine-style zoom against its B4 mount ENG equivalent.
Have you found you get enough IR contamination in bright exteriors to need an IR filter?

[Bob Hart]

Mark.


The B4-mount is P+S's own IMS-B4 adaptor with a thick correcting piece of prism glass in it. I sort of had my doubts but it does work well. I had tried the TV-Nikon direct and it was really soft. Wht the prism glass it sharpened up considerably.

As for IR contamination, I had not really looked for it closely but I have found on one day only, there was a bit of a false colour, yellowish versus orange for laterite rock over here in really bright clear-sky November daylight. It was that sort of daylight where you get that cold burn after about 15 minutes, which is UV rather than IR, so I am not sure what was going on or indeed if anything was.

It seems to diminish when you crush the blacks a bit in post, so it might just be the expanded dynamic range or maybe I got the whitebalance wrong.

The TV-Nikon lens is the more or less standard arrangement of a 8mm wide-end to 125mm tele end with a doubler to move the tele end up to 250mm and a macro function which shares a helicoid with the backfocus adjustment. From vague memory, it is fairly fast, f1.5 or f1.8.

For roaming about it is convenient, like the news cameras for which it was originally made. The servos on this one are cactus but the SI2K will apparently power a working ENG lens from the socket on the front of the body.

You can't use the lens run button though. You apparently have to use the SI2K remote button on its own handle. We are thinking of isolating the lens run button from its circuit and patching to the SI2K's own separate remote port when we can get the pinout diagram for the plug on the SI2K.

The TV-Nikon lens covers the full sensor at 2048 x 1152 on Steve Rice's camera, but there is a very slight vignette on the left corners of mine. If you look closely at the river images on this post, you may observe it.

http://www.dvinfo.net/forum/silicon-...-eng-lens.html

It is most evident at wide, the it disappaears, then re-appears part way through the zoom at about 20mm or so then is gone for the rest of the zoom movement and is not evident at all with the doubler. The Angenieux 10mm - 150mm zoom also does something like this.