My new transfer units: update

[MovieStuff]

In fact, we are experimenting with the idea of modulating the LED during the scan to control contrast. The idea is that we change the PMW as the camera scans across the image, lowering the intensity of the exposure when the camera is crossing highlite areas and increasing it in darker areas. Roger

Will this not introduce flickering? Sometimes there is a big difference in intensity between frames from the same scene.

Our experiments show that using this technique actually equalizes those tiny variations. I mean, that's the whole idea; to be able to make inter-frame changes in density during the scanning process. The circuit is pretty finnicky, though. Not sure if we're going to use it in a release version or not. But when it works, it beats the pants off of multiexposure technique. You can actually turn a knob and see the detail in the black area come up while the midtones and highlites stay the same.

Roger

[VideoFred]

But when it works, it beats the pants off of multiexposure technique. You can actually turn a knob and see the detail in the black area come up while the midtones and highlites stay the same.

Roger

Build it!
Sell it!
I'l buy it!

I'm looking for something like this for two years, now.
But does it work with entire frames or certain areas from a frame?
I mean, how can you bring up detail in dark parts without blowing out the highlites?

It must give feedback to the camera settings too, right?

Fred.

[MovieStuff]

I'm looking for something like this for two years, now.
But does it work with entire frames or certain areas from a frame?
I mean, how can you bring up detail in dark parts without blowing out the highlites?

It must give feedback to the camera settings too, right?

Well, yeah. That's how it works. A basic PWM system takes a reading from, say, the peak levels of the camera and adjusts the LED output to correct exposure. But the camera scans across in lines and not the whole picture at once. So, if the LED modulation is fast enough, you can change the intensity of the backlight as the camera scans, making changes to the intensity as the camera scans across different regions of the film frame. In other words, as the camera is scanning across a white area, the intensity would be lowered but as the scan of the camera crossed into a darker area the intensity would increase. It happens inter-frame. That is how you can increase brightness in the darker areas while keeping the highlights from blowing out because they are actually exposed at to different levels during one single scan.

Roger

[VideoFred]

Roger,

I admit: this is a genious idea :!:

You are following the camera scanning process with the backlight!
This must be very fast of cource, but this is not a problem with electronics.

But if the camera scans in lines, and a white spot is in the middle of a frame/line, what happens with a possible darker area at the left and at the right of the white spot?

Or do you follow the lines, to? Then it is almost point by point, pixel by pixel. Then correction must be realy damn fast :twisted:

This needs special software, right?
And could this work with any camera?
Could this function be added to, say, Cinecap (to give an example )

Fred.

[paulcotto]

I think we discussed the possibility of using an LCD monitor to do kind of what Roger is talking about. I was thinking of feeding just the luma signal to an LCD and inverting it to negative, then use this to back light the frame being transferred while using a proc amp to adjust the gamma and luma levels to brighten overly dark areas and reduce luma to blown out areas. You would need to use a lens to reduce the LCD output to the size of the frame and also a way to move the LCD to align it to the gate.

Regards,
Paul Cotto

[Patrick]

Roger, this sounds like quite innovative work with LEDs. Hmmm...with regards to functions found in broadcast cameras, I wonder how effective features like Knee and Master Pedestal are in minimising contrast.

[MovieStuff]

Or do you follow the lines, to? Then it is almost point by point, pixel by pixel.

Exactly.

Roger

[MovieStuff]

Roger, this sounds like quite innovative work with LEDs. Hmmm...with regards to functions found in broadcast cameras, I wonder how effective features like Knee and Master Pedestal are in minimising contrast.

They can do a lot to control contrast but are generally a "preset" that affects everything the camera points at, whether you need that adjustment or not. This would allow the camera to remain set up at standard factory specs and then modulate the image to obtain the contrast desired.

Roger

[christoph]

Or do you follow the lines, to? Then it is almost point by point, pixel by pixel.

Exactly.

well, i guess i just dont like all things "auto", but a subtle on the fly adjustment for very dark and very bright areas could certainly be useful for high volume work.

problem is, you'd need a good "curve" to control the amount of correction, oterwise the image will look unnatural.
other problem is, the circuit has to be fast. really fast. assuming that we have a camera that runs on 50hz (pal) and about 500 vertical lines, that makes 1/25,000 sec per line.. 700 pixels per line, thats 1/17,500,000 sec per pixel which means you'd need a PMW with about 50 Mhz if we allow for some overhead. not sure if a white high-power LED can react that fast, as they use phosphor to get mre efficency out of the LED. andyway, the whole circuit would be able to react withing nano seconds and even then, it can only react to a signal that s already there (ie too dark) so it could correct for the next one, but i dont see how it could react on the current one (this could actually result in a sharpening side effect).
and then you still have the problem of getting a really good color spectrum with LEDs (even though most people wont mind).

but a lot of things work in real world that in theory sound dodgy, so if you say you had some promising experimental results it's worth a try. i seriously doubt that it will work better than a proper multi capture... but then again, it certainly will be about 50 times faster. ;)

++ christoph ++

[MovieStuff]

but a lot of things work in real world that in theory sound dodgy, so if you say you had some promising experimental results it's worth a try.

Reaction time is obviously an issue that has to be dealt with or you get a halo effect, kind of like old black and white television from the 50's. That can be minimized by adjusting its threshold but what works well for one gamma situation doesn't always work well for another. You can set it for, say, someone wearing a white T-shirt in the bright sun standing against a dark background but then if it cuts to inside with more muted tones, it freaks out and over-reacts. As I said before, it is very finicky, which is why we have not pursued it more aggressively. Even if we got it to work reliably, the price of the final product would have to reflect the man-hours invested in the development. I don't think anyone could afford it on a practical basis.

i seriously doubt that it will work better than a proper multi capture...

"Work better" is relative. The end result can certainly look better (when it works!) because you never have a potential of "blooming" on highlights due to overexposure of the darker areas. That is an issue that can sometimes happen with the multi-capture approach. The modulated LED method avoids it because you can "feather" the dither of the exposure when going from dark to light rapidly; kind of like a soft-edge matte. We did it to test the theory. It can work, which is fun, but I fear it would take way too much R&D to make it viable.

How is your multi-capture system going?

Roger

[christoph]

i was away for a week but i'm capturing a test roll justin sent me as we speak (err... type ;)

at the moment i do 3 captures per frame (simply because my projector has three blades and i was too lazy to make an extra wheel for the light barrier trigger) but since it's a negative i dont need an extended range so i average all three to get less base noise.
the cameras dynamic range can nearly handle reversal as well, but i guess i'll do 2 passes with more light for the shadows and one pass for the highlights later on..

but next plan is to get a faster compuer, and everything wired over gigabit ... cos one 50ft rollresults in about 100GB of temporary data - urck

i'll post a few frames here if it works out allright (and if justin approves)
++ c.

[paulcotto]

I wonder if you had 3 different size blades Small, Medium, and Large if this would vary the ammount of light enough to automate the 3 exposeures? Or what if you put 3 progressively dark ND filters between the existing blades.

Paul Cotto

i was away for a week but i'm capturing a test roll justin sent me as we speak (err... type ;)

at the moment i do 3 captures per frame (simply because my projector has three blades and i was too lazy to make an extra wheel for the light barrier trigger) but since it's a negative i dont need an extended range so i average all three to get less base noise.
the cameras dynamic range can nearly handle reversal as well, but i guess i'll do 2 passes with more light for the shadows and one pass for the highlights later on..

but next plan is to get a faster compuer, and everything wired over gigabit ... cos one 50ft rollresults in about 100GB of temporary data - urck

i'll post a few frames here if it works out allright (and if justin approves)
++ c.

[christoph]

I wonder if you had 3 different size blades Small, Medium, and Large if this would vary the ammount of light enough to automate the 3 exposeures? Or what if you put 3 progressively dark ND filters between the existing blades.

well, the exposure is controlled by the shutter of the camera so the size of the blades doesnt really matter... but the idea with the ND filter is exactly what i'm planning to do (although i probably only use one or two filters)

++ c.

[Patrick]

"...are generally a "preset" that affects everything the camera points at, whether you need that adjustment or not."

I know not all the details in the manual for the DXC-950 are very specific but from reading about the Master Pedestal option, I get the assumption that it only lightens the dark areas of the image, leaving the rest of the image unaffected. I would like to see if this really is the case when I get the camera operational for testing. M. Pedestal is adjustable in increments from -99 to +99 which sounds quite versatile.

[christoph]

I know not all the details in the manual for the DXC-950 are very specific but from reading about the Master Pedestal option, I get the assumption that it only lightens the dark areas of the image, leaving the rest of the image unaffected.

the problem is that the CCD chip itself has only a limited range, no matter what you do with you master black, knee, or any other processing option that your camera has.
if a pixel on the chip doesnt get enough light, all you see is dark noise, if it gets too much light, it overfills and in worst case starts to bloom into nearby pixels.
so while good signal processing certainly is better than a bad one, the only option for true extended range really is to get a camera with a good chip.

failing that (as they are damn expensive) you might try variable backlight or multi-captures as discussed in various threads
++ christoph