First scans

[Paul Bruening]

Hello Gang,

 

I still have much tweaking to do on this scan rig. But, here's an early scan test. It's about 4.5K wide from a 2-perf negative image (knocked down for the web, of course). The neg was really thin so it's pretty grainy.

 

The second image is a close up. I'm having trouble getting around the blue spots. Working in high-res TIFFS helped. What's a good trick to massage out the blue splotches? They appear to be a part of the dye clusters on the neg (keeping in mind that they're reverses of the neg).

[Bruce Taylor]

Paul,

I know nothing about scanning, but I am very impressed that you have your rig up and it works. Go 2 perf go!

 

Bruce Taylor

www.Indi35.com

[Paul Bruening]

Still tweaking. But, here are some more:

[James Steven Beverly]

Just outta curiosity, are these shots of you, Paul?

[Phil Rhodes]

The blue splotches (and the red splotches and the green splotches) are grain - I'm not really sure what you expect to do about them! You could scan it on a Spirit and they'd still be there.

 

P

[Paul Bruening]

Yea, Steven, that's my chimp's-butt mug. I had Bruce, G'day, McNaughton install a radio remote controller on the camera. So I can start and stop from as far away as 100 yards (that's the small controller that I'm using in my left hand). With a video transmitter, I can watch the camera's image on a TFT-TV and start and stop the camera all from a distance. That's for the days when my deferred pay, hung-over crew doesn't show up and I have to run camera as well as direct.

Phil,

 

Am I resolving about as well as can be expected? What is your opinion of that close-up (keeping in mind it is a thin negative)? My concern is that the blue splotches in particular have too strong a presence on the image.

[Paul Bruening]

Hey Phil,

 

I was thinking something along the lines of a luminosity, digital matte that isolated just that range of irksome blue splotches. Then do 33% transparency, normal blend to tone the color down a little. I don't mind them being there, obviously. I do mind how strong they stand out. What do you think?

[Phil Rhodes]

Well, it's tough to tell without seeing a comparison with output of the same frame from another device, but I've seen scans with that sort of grain from all kinds of things. It's really really JPEG'd, of course, but that could just be the way you've presented it for the web.

 

I'm sure you could come up with all kinds of interesting methods for reducing the apparency of grain without reducing apparent sharpness of real detail - a lot of people do!

 

P

[John Butler]

Could you direct me to information as to what was used on this telecine? (you imply its a diy)

[Paul Bruening]

I don't know of any information I can direct you to. I could never find anything on the web about how to do it. There are some old threads here from when a bunch of us were hacking over ways to do a cheap DIY DI. I can tell you what I did. But, I can't guarantee you that it's the better way to go since I'm still having troubles with the system.

 

I had a Mitchell NC converted into a projector by Bruce McNaughton at Aranda Group in Australia. It was difficult to do for 2 perf but I'm thinking it is not possible for 4-perf because of clearance issues behind the pressure plate. Then, I shoot into the gate with a DSLR and a macro lens. Paul Moorcroft made a GUI utility via a computer to drive the stepper motor on the NC and take the snaps with the DSLR. I'm building a 3 way, linear stage for the solid mounts. That's about it... if I could just iron out the kinks.

[K Borowski]

Hey Paul, that's a sexy light meter you have there ;)

 

I notice that there seems to be some color irregularities in the frames you've posted, sort of like irregular illumination that is backlighting the film when you photograph it.

 

Do you have your light diffused? If not you should. What's the color temperature of the light source? Do you have your camera color balanced to the correct temperature?

 

What do you have your DSLR mounted on the insure image stability from frame to frame?

 

~Karl

[Paul Bruening]

"Hey Paul, that's a sexy light meter you have there wink.gif"

 

Hey Karl. My wife says it's too big. Wait... what are we talking about???

 

"I notice that there seems to be some color irregularities in the frames you've posted, sort of like irregular illumination that is backlighting the film when you photograph it."

 

Yea, I didn't standardize the inverting procedures between these test frames. So, they look inconsistent one to another. The settings and steps of inversion will be established in preset macros that will deliver a sameness between frames in a sequence. I'll set up a catalog of macros to apply to the most common shooting situations and emulsions. It is unavoidable that that means I'll be pre-color timing the images.

 

There is an inconsistent range of light values and color in the white background element. I'm working on the best way to compensate for that digitally. Currently, I'm using an inverted exposure of the background element in a 30% transparency layer blend. I'm thinking there's a better way and am still mucking around the menu bar.

 

These exposures include blockages of light due to the viewing angle of that lens and the transmission angle of the backlight. That's the band of lighter image along the top of the frame (darker band on the negative due to the blocked light). I've ground out some of the pressure plate opening to eliminate the blockages.

 

Something I never accounted for is the irregularity of light caused by the taking lens. There is about a 1/3 stop or less circular gradation of light that occurs in the middle of the image and diminishes out toward the edge. It appears to be a phenomena of the iris in the taking lens. There is no evidence of this phenomena when only the BG of the NC is shot. It happens in varying degrees depending on the taking lens. I shot this neg wide open using ND filters on every tested lens. I'm thinking that that may have imposed this inconsistent value gradation on the negative. Either way I can simply overlay an inverse image and transparency it in to equalize the values.

 

"Do you have your light diffused? If not you should. What's the color temperature of the light source? Do you have your camera color balanced to the correct temperature?"

 

I can white balance against the white BG or preselect from a range of color balance settings provided by the DSLR contol software.

 

"What do you have your DSLR mounted on the insure image stability from frame to frame?"

 

I'm building a rail and stage system out of steel (sort of like on a lathe). The NC has a one way stage. The DSLR will have a 3 way stage with lockable gimbal for fine tuning. I'm using 1/2" X 12" X 12" steel plate for weight and stability on the camera mounts. I would rather have a cast and machined stage system but I'm very, very broke these days. The steel I got out of the discards pile at the local supply house. I already had the MIG welder and cut tools.

[K Borowski]

"I notice that there seems to be some color irregularities in the frames you've posted, sort of like irregular illumination that is backlighting the film when you photograph it."

 

Yea, I didn't standardize the inverting procedures between these test frames. So, they look inconsistent one to another. The settings and steps of inversion will be established in preset macros that will deliver a sameness between frames in a sequence. I'll set up a catalog of macros to apply to the most common shooting situations and emulsions. It is unavoidable that that means I'll be pre-color timing the images.

 

There is an inconsistent range of light values and color in the white background element. I'm working on the best way to compensate for that digitally. Currently, I'm using an inverted exposure of the background element in a 30% transparency layer blend. I'm thinking there's a better way and am still mucking around the menu bar.

 

These exposures include blockages of light due to the viewing angle of that lens and the transmission angle of the backlight. That's the band of lighter image along the top of the frame (darker band on the negative due to the blocked light). I've ground out some of the pressure plate opening to eliminate the blockages.

 

Something I never accounted for is the irregularity of light caused by the taking lens. There is about a 1/3 stop or less circular gradation of light that occurs in the middle of the image and diminishes out toward the edge. It appears to be a phenomena of the iris in the taking lens. There is no evidence of this phenomena when only the BG of the NC is shot. It happens in varying degrees depending on the taking lens. I shot this neg wide open using ND filters on every tested lens. I'm thinking that that may have imposed this inconsistent value gradation on the negative. Either way I can simply overlay an inverse image and transparency it in to equalize the values.

 

"Do you have your light diffused? If not you should. What's the color temperature of the light source? Do you have your camera color balanced to the correct temperature?"

 

I can white balance against the white BG or preselect from a range of color balance settings provided by the DSLR contol software.

 

OK, I don't think you understood my question. I'm talking about inconsistencies within individual frames. It looks like you have color shifts through a frame that seem to be caused by uneven illumination.

 

You do NOT need to overly complicate things by fixing them in photoshop. You need to make sure your illumination is consistent behind your negative edge to edge. Use a light meter, densitometer, or some other light comparitive device to check this.

 

If you have a lens that is giving you irregular exposure on the film, jettison. This is a TEST and you MUST eliminate variables or else you're just burning money on your front, err here it'd be back, lawn.

 

I also do not think the noise you're getting is grain, it might be caused by grain, but I suspect it is grain aliasing. Do you have any control over image sharpening in your DSLR? AFTER you fix your illumination problem and your lens problem, I'd recommend dealing with that.

 

As far as the practicallity of this whole thing goes, is it realistic to use this setup for an entire film? What sort of time per minute of footage do you anticipate? Obviously you can't transfer in real time with a DSLR. I don't know of any digital camera that can shoot uncompressed images at 24 fps.

[Paul Bruening]

No real hurry on the scans. How about $0.005 per scan? Is that reason enough to go this route?

[K Borowski]

No real hurry on the scans. How about $0.005 per scan? Is that reason enough to go this route?

 

It's a good reason, but even at $.0005 per scan, if the scanner's no good, you're getting what you're paying for.

 

There's significant progress that has to be made here.

 

Don't forget, a shutter is only rated to 100,000 cycles on an SLR. That's a 70 minute movie right there at 24 fps. So you're talking $150-200 right there.

 

You need to even out your illumination, color balance your camera (it looks like it's not balanced and you're trying to adjust later, which will not look as good as shooting with the correct color balance to begin with; trust me I learned this the hard way).

 

You also need to make sure your camera is dealing with pixelation. What ISO are you shooting at?

 

~Karl

[Marc Roessler]

Nice for a start

 

I don't think the blue clusters are grain, either. If they were, that stuff would pop up in the contact prints as well... have one made and compare. Film grain looks different.

 

As a side note.. Some folks take a look at S16 or 35 scans, with super duper magnification, and then they start complaining "oh look at all this crappy film grain" - when the "grain" does actually originate in the scanner, as electronic noise! The magic moment for me was when I had an image from a first 35mm print (inluding soundtrack) scanned: there was horrible "grain" in the analog optical sound track (!) - go figure.

 

Regards, Marc.

Edited June 8, 2008 by Marc Roessler

[David Mullen ASC]

It would be better to shoot a properly exposed MacBeth chart using fresh stock to reduce the variables of your scan. To me, the blue noise/grain looks like outdated stock perhaps underexposed. Either that or your color-correction approach is adding a lot of noise to the blue channel.

[Paul Bruening]

To me, the blue noise/grain looks like outdated stock perhaps underexposed.

 

You have called it perfectly. It was old stock and way underexposed.

 

Even though I am concerned about the quality of my scanning technology, I kinda' like the look of these neg/digital files from the test roll. It has a sort-of Apocalypse Now look. I'm sure A.N. was properly exposed on good stock. I just remember the prints I saw from long ago as seeming a little gritty and undersaturated.

 

I was going to roll TAF/TEC strips through the system, both telecine and scans to get it all waveformed and vector scoped up to a reasonable ballpark of standards. I figured, then, I'd do Macbeth and other charts to see how the system handled real world exposures once standardized to TAF/TEC. I'm assuming that Kodak's TAF and TEC are the very best equipment grading sources to start the whole process.

[Sean Morris]

G'Day Paul,

 

How are you dealing with the orange mask from the neg to a digital file, a custom curve?... just out of interest.

 

Cheers

Sean

[Paul Bruening]

In this time that I'm getting the system up and running, I'm simply using the "Invert" and "Auto Levels" commands in Photoshop. Then, I tweak the image with the usual "Image" commands. When I get the system automated, I'll likely try to employ somebody's software negative inverter. There are utilities here and there that do this. I haven't done all the research yet. I've been involved with the hardware concerns, lately. If I can't find a fast enough inverter, I'll just macro the process in Photoshop or even After Effects. The data has to be pretty heavily crunched, dust busting in Digital Ice anyway. A little extra time inverting images won't really get noticed.

[Guest will griffith]

In this time that I'm getting the system up and running, I'm simply using the "Invert" and "Auto Levels" commands in Photoshop. Then, I tweak the image with the usual "Image" commands.

 

Have you considered using RAW?

 

You can open the "scans" in Adobe Bridge, copy and paste

your raw settings (curves, WB, exposure, etc. etc.)

to any number of frames and then open the RAW image

sequence in After Effects for heavier editing.

 

I would think shooting RAW would save a lot of time,

make things more flexible, and your future clients wold

also appreciate the ability to adjust the original scan(s)

without actually modifying the files.

 

Also... how are you timing the recordings? Can you set it up so

each time the frame advances it clicks something and fires the

camera from the PC/Mac such as Canon or Nikon's software allows

you to do?

 

You could set it up to auto advance and go take a nap by your pool

while it works for you. :)

[Phil Rhodes]

I have fiddled around with this sort of idea myself, and came rapidly to the conclusion that DSLR lenses really aren't set up for it. The lens I used, the Canon EF-S 60mm macro prime, is a pretty decent piece of glass by all accounts but flares rather badly when used at the rather extreme settings this sort of thing really requires. I think you may have the same problem.

 

P

[Paul Bruening]

Hey Phil,

 

I'm using the Canon MP-E 65mm 5:1 Macro (Micro) lens. It's a metal barreled lens, not one of the crap-ola plastic lenses made by Canon these days. Flare hasn't shown up as a problem as far as I can tell. If you can see something I don't, please, point it out to me. My back light isn't that bright. It's a 100W, 12V halogen bouncing off of two surfaces before going through the negative. I'm still toying with the idea of using a polarizer to cut stray light. It's on the list of "Things I should test out."

 

Will,

 

The DSLR captures 5K RAW and JPEG proxies simultaneously and sequentially to hard drive via fire wire. The projector is run by a software utility provided by Paul Moorcroft via Bruce "G'Day" McNaughton. It links to the Kodak camera driver. The rig's GUI coordinates the DSLR and projector so I can set the shot list and then go f*** off until the roll's done. One of the things I like about Bruce's design is that the projector single frame advances through each scan-sequence of shots but rolls at 24fps through the bits between scan frames. The computer I'm using for capture control and storage is a 64bit dual core. If I can scratch up the cheese I'll bump it up to quad core. It has 6 SATAII (3.00 G/s) plug points on the MoBo. 1TB SATAIIs are getting below $200.00 each. The capture rate is pretty solid even though the Kodak DSLR is considered a little slow on its preprocessing.

 

I find that TIFF files hold up well to digital manipulation but I'm sure I'll have to stay RAW or bust them into DPX files since that's still the movie industry standard. I'll decide between results, speed and any customer (IF any) expectations a little farther down this road.

[Phil Rhodes]

I dunno, I'm just struck with the feeling your images might be subject perhaps to a little of what might be better called veiling. Might just be the grade.

 

Certainly you could pay for worse off some shot old Rank 3.

 

P

[Guest will griffith]

The DSLR captures 5K RAW and JPEG proxies simultaneously and sequentially to hard drive via fire wire. The projector is run by a software utility provided by Paul Moorcroft via Bruce "G'Day" McNaughton. It links to the Kodak camera driver.

Sounds like you have things pretty well covered.

Now you just need some flashy marketing materials. :)