Sexy models shot on Super 8 in Cuba - 100D - 2K Transfer

[Justin Lovell]

Shot by cinematographer Pouyan Fard with a cheap-o super 8 camera picked up for $3 at value village. Will be Intercut for a Fashion commercial spot with his 5D DSLR steadicam shots.

This will definitely stand apart from the DSLR footage. The texture and softness of super 8 really works well to get a less crisp/harsh dreamy reality. Does wonders to women's skin tones.




Transferred in 2K at http://www.framediscreet.com.
Processed at http://www.NiagaraCustomLab.com

Originally posted here (incase the board crops the images)
http://www.framediscreet.com/about

[carllooper]

Nice.

[carllooper]

Hi Justin,

what are your prices for a direct 2K RAW capture of 8 and 16. Your website doesn't list prices for this. I assume this is because your clients don't usually have the facilities to deal with RAW files, which is understandable. For me, though, I write image processing software (amongst other things) so have no problem parsing RAW files - so long as the file spec is available. In other words I don't need any supervised transfer - just RAW data. I've been looking for someone who can do 2K RAW transfers (or better) for some time now.

Can you tell me whether you use a single chip sensor or a three chip sensor. As you no doubt know, a single chip 2K sensor would have a bayer filter, meaning the true scanning resolution through such would be at best 1K (for green) and less for blue and red. Debayering algorithms improve this considerably of course and I've written a few such algorithms myself, but they interfere with particular information I'm after in my own work.

Carl

[Nicholas Kovats]

Carl,

Justin has been scanning my Super 8 footage since 2006 on a WorkPrinter system. His and Lianna's "real time" color correction of my reversal and negative footage has been superb. He has since upgraded to a custom sprocket based 2K rig which utilizes a Silicon Imaging SI-2K camera head, i.e.

http://www.siliconimaging.com/DigitalCi ... tures.html

They recently performed a scan of my UP8 footage with the new system. Deliverable were AVI + MOV not RAW. I will try to upload a sample frame with the associated specs.

Nicholas

[carllooper]

He has since upgraded to a custom sprocket based 2K rig which utilizes a Silicon Imaging SI-2K camera head, i.e. http://www.siliconimaging.com/DigitalCi ... tures.html

Hi Nicholas,

thanks for that. I notice the SI-2K is a single chip sensor which is really the problem I need to avoid. While it obviously gives very good results such sensors actually interfere with the algorithms I've been developing. Basically such sensors sample the green channel through every second sensor pixel, with red and blue being sampled though the other (first red, then blue, etc).

Normally this doesn't matter. Debayering algorithms reconstruct missing colour pixels from the existing ones. And a scanner typically does that debayering (or another process following a scan). And it looks great. But this debayering actually interferes with what my super-resolution algorithms do. Or to put it another way they reduce the effectiveness of what is otherwise possible to do.

That said I've been working with bayered scans and while not optimum they still work quite fine - just not as fine I'm hoping to achieve. So more than happy to run any images you have through what I've got. Will be cool.

I'll set up a dropbox folder and email you. They'll need to uncompressed files, eg. png files. Given the file size of such don't drop too many in. A few seconds worth (eg. 48 frames) to start with.

Ok. You should recieve an email with details about how to access my hard drive. Basically you install dropbox which creates a folder on your hard drive which mirrors the one on my hard drive. You drop files into your folder and they appear in mine.

I've called the folder "UltraPan8".

ps. you don't have to install DropBox. You can also use the web interface to upload files which will then appear on my hard drive.

[Justin Lovell]

Hey Carl!

Sounds like your post processing setup is pretty crucial to many factors. Currently 2k RAW scans are priced the same as best light 1920x1080 scans at $30/roll.

It's taken me a while to get a setup developed that I am happy with. Spent well over a year in R&D with different sensors and engineers to get it to this point.

Our 8/super 8 system is spocketless. Our 16mm belladonna system is sprocket driven.
While Nick knows a bit about our system, he doesn't know all our secrets ;)

Thanks for your interest!

[carllooper]

Sounds like your post processing setup is pretty crucial to many factors.

Thanks Justin. Yes, it's something I've been working on for more than a year now. And I still haven't fully exhausted the potential of the processes. My only limitation at the moment (apart from implementing better algorithms) is getting higher definition scans. That doesn't mean I'm getting bad results from available scans. On the contrary I'm getting excellent results. But I know there is even better results I can get. Conventional wisdom is that there is a limit beyond which the only "information" you'll get out of increasingly higher definition scans won't be any more of the signal - just more noise. But I'm not seeing that. I'm still not seeing where the signal limit might be. Indeed, apart from limits of a lens (which even there I wonder) I don't know if one can really say there is a limit.

Indeed most of the literature speaks of subjectivity as ruling the boundary. Another way of saying that is the limit is actually undefined. Subjectivity is required to rule on where that limit might be because the mathematics/statistics/instruments don't know!

Carl

[carllooper]

[Justin Lovell]

Cool. Nice correction. The cinematographer, Pouyan Fard, was looking for some more contrast so I punched it up a little. You should see what the RAW looked like from our scan, even more range to work with than what you had there!

[carllooper]

Cool. Nice correction. The cinematographer, Pouyan Fard, was looking for some more contrast so I punched it up a little. You should see what the RAW looked like from our scan, even more range to work with than what you had there!

Thanks Justin. It's a pleasure playing with film scans, even on just jpegs of those scans. As you suggest - imagine what's possible with the RAW files. The grain of film really helps in tuning the image. As you push/pull any given area of the image, one pixel might hit the boundary of the image, but the one right next to it doesn't, staying within bounds. The same occurs at the other end of the range. You can't do that with a digitally acquired image. You could try adding grain to a digital image but it doesn't work. Unlike digitally added grain, the native grain of film is statistically correlated with the content of the original signal. It's very pleasant. Very flexible.

For motion shots I use motion tracking to move area specific grading decisions in a content aware manner - in case anyone is wondering. A lot more work to get right. But correspondingly more pleasure when achieved.

[Nicholas Kovats]

Very impressive!