Except for the P+S Technik Pro35 and a relay path via the special IMS-B4 lens mount adaptor, it seems there is no other option for the SI2K camera.

There is also no facility with the Pro35 to change the relay lens view of the groundglass (area scanned) when lower pixel counts are selected, ie.,1280 x 720. Changing the pixel count has the collateral effect of reducing the active sensor area, which in turn reduces the area of the groundglass screen "seen" by the camera.

Therefore shooting a groundglass at 1280 x 720 introduces two resolution hits if the relat field-of-view cannot be widened, that of fewer pixels per image frame and the "scale" size of each groundglass "grain" being larger relative to the size of the image frame.

By interesting co-incidence, 5 microns is apparently the optimum grade of abrasive for groundglass texture dressing - and the size of individual pixels in the camera sensor.

Groundglass imaging has its naysayers, apparent inferior resolution performance in low light, risk of artifacts and the inconvenience of managing two optical stages, being the significant complaints. It remains however the only true "non-coherent" process of real-time motion imaging in the electronic realm.

If a flexibly relayed 35mm groundglass adaptor for the SI2K, capable of the practical resolution of the SI2K at a 1920 x 1080 setting became available, are there any owners who would be interested?

Why?

To compress DOF (if so at all needed...), just use a telephoto lens...

or, use a ND filter in front of your lens and have it wide open...

Not enough? - why not use your Canon 7D or 5MII as 2nd cam for that super shallow DOF shot... Both have larger than Super 35mm sensor sizes...

In general, I actually find SI2K's DOF even too shallow for general shots...

Alex.


Why indeed? Probably the luddite in me resistant to change is to blame.

I don't have a 7D or 5D.

I do have a 35mm groundglass device which I have done some mods to. I have been getting some good looks even with the hacks I have made. I get the SI2K's dynamic range and no sharpness loss due to compression which occurs with prosumer cams as the cineform codec is more robust.

The SI2K is also very resistant to moire artifacts. The viewing system will warn you about risky subjects like fine patterned fabrics yet the camera will reproduce them without problem.


A possible relay solution already tailored to single CMOS sensors from an existing vendor may become available. I understand it would require a minimum production run of about 25 units to be break-even viable.

The same relay may work for tube based passive IR. European sourced tubes are now somewhere between SD and HD resolution which would be handy for documentaries. That has not been tested for.

It would also be handy for 35mm reversal and neg stills scanning via the SI2K, with ability for subtle framing adjustments. That is something you can already do with other lenses and dioptres but with a lot more messing arround to get right. Or you can buy a stills transparency scanner.

I agree that for most purposes, the available depths-of-field from fast Super16mm film lenses are adequate. Extremely shallow depths of field can be had when wanted with lenses longer than 25mm. However those come with the signature penalty of all 2/3" imaging systems, a perspective distortion of human faces.

A DP in Greece thought groundglass imaging was worth going for with another 2/3" HD system and the SGPro 35mm adaptor. He achieved acceptable corner-to-corner sharpness with Zeiss primes and custom barrels.

I am fishing because I may be able to get hold of the manufacturer's prototype for practical testing for six weeks.


There is of course unbridled and corrupt self-interest motivating my intial post.

If this relay and adaptor is a good solution, I would like to have the option as would the other SI2K operator over here, However it will not become affordable unless sufficient SI2K operators also want it, hence my enquiry here.

It might also not happen except by a pre-order and deposit scheme because of the same short production run and very tight margin of risk between breakeven and loss if stock is unsold.

Bob: "Extremely shallow depths of field can be had when wanted with lenses longer than 25mm. However those come with the signature penalty of all 2/3" imaging systems, a perspective distortion of human faces. "

Alex: hmm.... I thought distortion occurs on the wide end, not telephoto... That's why in photography we prefer telephoto lenses in portrait settings...

Or did I misunderstand the statement? (in which case I'm sorry...)

Speaking of Canon 7D: just shot an entire comedy short (3min edited, it's my own web site promo with a twist...) with it, see it here (http://www.youtube.com/primehdtube). That camera is banging, yo!

You can only change perspective by changing the camera position, so I'm not sure about this point. Although, I would say that the f1.4 35mm Zeiss Contax lens that I had for my 16mm Aaton displayed barrel distortion, whereas the T1.3 25mm Super Speed designed for 16mm had no visible distortion.

Brian. thanks for making that point. ( This is where the real practitioners get sorted from the jack-of-all-traders like me.)

The word "perspective" is probably a misuse by me, prompted in part by perspective technical drawing which was one of my niche skills in my school years, alas now made muchly obsolete with CAD programs and printers.

Alex. What I refer to is the foreshortening or flattening effect along the optical centerline on objects when a lens which confers a narrow angle-of-view is chosen. A wide lens tends to stretch the face towards and away from the camera.

This is distinct from distortion which is stretching of the image itself in directions 90degrees relative to the optical centerline.

To be honest I've never noticed these effects when shooting with 2/3" other than variations of quality between different 2/3" lenses eg the Canon wide angle zoom and their standard ENG zooms.

At 20/25mm 2/3" doesn't look that different to a 50mm on a 35mm film camera (apart from DOF), the subject camera distances for the same framing, from memory, aren't hugely different.



I

Brian.


I and most would agree with you.

I ran into a rare exception shooting a light aircraft cabin interior, when forced to using a 58mm for some close-ups and to defocus the closer background. This was because of confined workspace, external airframe structure and faking the under-wing background with whiteboards to look like burned out sky. The vision whilst fine, made the actors look un-naturally fatter in the face.

The 105mm or 135mm for a 35mm stills frame does not seem to cause that effect quite as profoundly.

I did make life hard myself by using a 5.7mm ultrawide for the in-cabin in-air shots. For practical and airsafety reasons, I could not shoot the enitre dialog whilst airbourne.

You do get variations in distortion between different lenses even at the same focal lengths.

Agreed.

The Noct-Nikkor 58mm apparently has a slight barrel distortion which might account for it.

Why oh why do you bother ? you may well ask.

Well - because I want to and I can already achieve good results on a EX1 and Letus Extreme.

For the moment, I shall keep the identity of the device, manufacturer and vendor under wraps until permitted to disclose but suffice to say, it is on its way via FEDEX.

It is a flip groundglass adaptor, rails kit and direct relay prototype, specifically engineered for the SI2K single CMOS imager system. If the groundglass area, texture and condenser optic are comparable with or better than the Letus Extreme, it should resolve 35mm groundglass relayed images at the practical resolution of the SI2K.

Myself and a local DP, Gavan O'Sullivan will get to experiment with it for about six weeks, put it up against the Canon 7D, hopefully subject it to all manner of creative lighting conditions and may make an already planned short film with it instead of on the bare SI2K.

Will fill up more of this space as more news comes to hand.

The custom SI2K prototype 2/3" single-sensor relay lens and 35mm adaptor arrived today, so will have a look at it all tonight and post how it fits up.

Very quick early test. - No centering adjustments were attempted. Poor lighting conditions existed for chart. Chart was not quite square-on. No corner softness or chromatic abberation was observed. Relay lens and 85mm f1.4 Nikon lens were wide-open for this test. Grounglass motor was not switched on for this test.

A couple of happysnaps from this afternoon's quick test. Relay focus was set by focusing sharp on groundglass texture, not on a test chart with groundglass running which is sometimes good for about an extra 50 TV lines of sharpness.

I forgot to reset colour temp to daylight so there is a blue cast to the images.

Here are a few more happy snaps with the zoom relay, a test chart under better lighting which improves things considerably and some pics of the rig with Letus Extreme and Brevis adaptors attached.

The Letus chart is not a fair representation as the motor was not running and the framing was slightly looser on the chart. The images were done in haste as I was recording direct to a thumb drive and had to keep the clips short.

What is apparent is that the prototype zoom relay for single chip 2/3" cameras works as well as any 35mm stills prime I have tried as a relay lens and does not soften the image corners. The achromat in the Letus was removed for the test.

The Noct-Nikkor prime lens used is noted in reviews as having a slight barrel distortion. Therefore, any such distortion observed in the images should be attributed to the Nikon lens, not the relay path.

Dennis and Hien, if you are around, here is a theortical query for you.

Have you been told any experiences relating to infra-red contamination of images when deep ND filtering is used with groundglass devices?

Mixed visible and IR sources have a very significant negative effect on tube-based night-vision as far as image clarity is concerned. My imagining is that the same might apply over the much longer optical path via adaptors and that maybe IR light-loss through the adaptor path may be also less than for visible light.

Bob, I've observed some very strange things with regard to ND filters and IR, particularly on our EX1 (another CMOS camera!). With the SI-2K, and Ari's advise regarding IR, I'm guessing there's a very good chance that something interesting is going on outdoors.

The chart looks good.

Cheers,
Dennis.

Now the flu is done and dusted and I am only half-dead, I thought I'd get one of the hard bits over first and try the relay in low light in a less well lit corner of the Armadale Retail dragstrip.

The relay was wide-open to provoke any softness it might have. The lenses were Noct-Nikkor 58mm f1.2 and Nikon 28mm f1.4, both backed off into a sweeter spot than wide-open. Camera gain was set to 0db. It is all about not being a fair test. I only shot 0db. I was getting some unwelcome attention from some people in a car across the road out of the shot. So I did not shoot at higher gains but packed up and left. It is a bit of a crimmy neighbourhood. I think they thought I might have been running a speed-camera, something I have been confronted with previously.

Here are a few grabs, two standard, two pushed in post. In the pushed grabs, I dropped the blue channel quite a lot. The camera white-balance was set to 3200K.

Given the hostile conditions, it seems to be resolving quite well.

I have just received the True-cut IR filter so will resume my tests with the Cinevate varifocal relay. I will also try it with the anti-aliasing filter removed as it is expected that the groundglass itself, with a notional 5 micron "grain" over an approx 27,000 micron groundglass screen width will perform that function.