[Konvas] DSLR telecine (off topic)

H. W. Stone colcam at aim.com
Wed Sep 3 11:42:17 CDT 2008


 Okay, I have to toss in two cents-- I'm not sure what that is adjusted for inflation,but--

You don't rip the camera apart.  If you want to digitalize a chunk of film it is easy.  You use a "zoom cropping" slide copier to transfer the image to the full DSLR frame, and put the slide copier into a holding block to mount to an indexable drive-- like one of the movements discussed before-- with a source balanced light source/diffuser behind it.  You bring up a frame, expose it, and then index the next frame and expose it.  It is really quite simple.  The point you need to realize is that the shutter and sensor for a DSLR are not intended for the amount of use cine requires, so IF you do much film you will need to change out bodies during your project.

24fps.

Twenty four frames per second.  

That's one thousand, four hundred and forty frames per minute of film.  

It means that a one hundred minute slug of film is beyond the maximum lifespan of the top notch DSLR sensor, let alone the shutter.

That was one of the bugaboos I ran into on my use of a DSLR sensor for cine capture, even when removed from behind the DSLR shutter and installed behind a different shutter, then slowed down to prevent overheating, they just did not design the DSLR sensor to last that many exposures.

Sorry.


 


 

-----Original Message-----
From: Sean McVeigh <konvas at smallpony.ca>
To: bradkonvas at yahoo.com; Konvas Discussion List <cinema at konvas.org>
Sent: Wed, 3 Se
p 2008 12:03 pm
Subject: Re: [Konvas] DSLR telecine (off topic)










I looked into this possibility about 2 years back...
You still want some sort of imaging element.. you can't "contact print" 
them since you need to keep the sensor quite isolated from the environment.
The sensors are fairly expensive, especially in the lower-defects 
grades, and you will need software to cover up the dead pixels / lines / 
etc.
The support electronics are not that complicated, but you need to do 
everything from scratch unless you are using an evaluation platform 
(again -- $$$$).  I believe I was looking at using CCD sensors, which 
didn't require much more than high-speed ADCs (more than one if your 
sensor can read out from different corners to improve throughput) and a 
CPU or even a powerful MCU.
I was hoping to find a good panchromatic sensor that would snap 3 RGB 
backlit images, but I don't remember getting very far with that, which 
was disappointing because all you need is about a 3MP sensor for some 
good 2K scans.  In the bayered sensors, you need more like 6MP.
A cheaper option was to use a linear sensor and sweep it over the image, 
which I believe is what the northlight units use.  Upside was that the 
sensors are about 1/5th the cost, and you can afford more resolution 
this way, and the support electronics are simpler.  Downside is that the 
mechanical assembly to precisely move the sensor in micrometer 
increments was quite expensive.  Might be a20better way to do this using 
some sort of drum or rotating sensor rather than a linear motor.

In the end, it seemed that you could find a good DSLR (like the canon 5D 
for example) for the same price or less of a good quality sensor, so I 
postponed the project until I reckon it's time to purchase a DSLR (which 
I still haven't done).  The only weakness with going the DSLR route is 
that damned mirror.. no one seems to provide a mechanical latch-up 
feature, whereby you could lock the mirror out of the way and take 
multiple (ie. thousands of) photos.  I always figured you'd need to take 
your camera apart and remove the mirror, since it is the component that 
is going to fail first.. and typically only after about 3 hours of 
footage is scanner.
But maybe cannibalizing a DSLR is just your sort of thing :)

Cheers,
Sean


Brad Leong wrote:
> A while back my dreams of making my own RED like camera were crushed… : )
>
> Had a new idea and wanted to get the konvas input to see if it’s less 
ridiculous – 
> Is there a way that I could fashion a makeshift yet high rez telecine machine 
using a DSLR sensor?  It wouldn’t be able to run at real time but it would 
hopefully be able to produce some high rez frame by frame scanning.  Thoughts?
>
> I don't know much about how telecine machines or film scanning works but it 
seems like sense the sensor CMOS size is equivalent you could build something 
rather easily
 that just advances the film over the sensor and snaps a picture 
with every advance.
>
> -Brad
>
>
>       
>
>
> _______________________________________________
>  



 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://konvas.org/pipermail/cinema_konvas.org/attachments/20080903/f2cea048/attachment.html 


More information about the Cinema mailing list