New 8mm scanner from Tobin

[retroman]

I dont believe what Kinetta writes .
Kodak announced some time ago, that regular8 and super-8 has a technical resolution of 700 lines
this would be 720p under laboratory situations.

So most of the homemovies are under and overexposed, so that they are not more than 500 lines.
If you compare regular8 to super8 , most of the regular have better colors and sharpness.

So Kinetta should has annouced regular8 with 3 K and Super-8 with 2K

retroman

[carllooper]

If you look at the technical resolution charts that Kodak provides (the MTF curves) they are incomplete. The printed curve typically stops around the 30% response level. This doesn't mean the response of the actual film itself abruptly stops at the 30% response line. The film itself continues responding, towards zero percent response at infinite resolution. However other factors, independent of film resolution, will intervene on what is meant by such resolution/response curves. For example, if Regular8 looks sharper over Super8 it can't be in terms of the filmstock (if using the same filmstock). It will be due to the more robust transport mechanisms in Regular8 cameras. But if using a Logmar Super8 camera the opposite result would obtain, due to it's pin registration mechanism and the fact that Super8 has a larger film area compared to Regular 8.

The reason the printed MTF curve for filmstocks doesn't continue down to zero is that you can just extrapolate the remainder of the curve from what is printed. Also the curves are designed for comparing filmstocks rather than characterising a particular filmstock in itself. The printed curve, despite being incomplete, does such a job.

Lets look at the printed MTF of an arbitrary filmstock such as Vision3 500T. The printed curves from Kodak, for each colour channel stop at about 80 cycles/mm, where the response for each channel differs for this particular resolution level. As already discussed this is not because each colour channel of the actual film itself suddenly stops at 80 cycles/mm. It's just an arbitrary cutoff chosen for how the film will be depicted in a graphic - in this case chosen in terms of a particular frequency rather than a particular response.

In film to digital transfers it's possible to enhance the otherwise low response of the film so that the otherwise undocumented response of the film to particular frequencies become more visible than they would otherwise be when watching the film itself. But to do that you need more sensor pixels. As many as you like or can afford. And appropriate digital processing to enhance such.

Counter-intuitively it is small gauge film that requires more sensor pixels to reconstruct it's signal when compared to a larger gauge film. Simply put, small gauge film requires more information be obtained from the film if one wants to reconstruct a signal with a "strength" that approaches that of a larger gauge film. There is greater interaction between the concepts of grain and resolution in the smaller film, so more information required to get "behind" such interaction.

C

[Tommy]

Overall, regular 8mm cameras had better quality lenses than super 8. My experience was that it was only the high end super 8 cameras that had high resolution glass lenses. Many of the mid to low end super 8 cameras had low res plastic lenses that actually had lower resolution than most reg 8mm cameras.

[carllooper]

Overall, regular 8mm cameras had better quality lenses than super 8. My experience was that it was only the high end super 8 cameras that had high resolution glass lenses. Many of the mid to low end super 8 cameras had low res plastic lenses that actually had lower resolution than most reg 8mm cameras.

Ah yes, that could be it. I've only had experience with Canon and Leicina Super8 cameras, where the lenses for those were very good. I've never encountered a plastic lens Super8 camera, but I take your word for it. Along the same lines, I imagine the lenses used on most Regular8 cameras would have been primes, whereas Super8 cameras had zooms (which would tend to lower the resolution a little).

Digital software is such these days that if the image is not too soft it can be brought back into focus using clever deconvolving processing ("smart sharpen"). I recall such being discussed by Lenny Lipton in the 70s (in the book Independent Filmmaking), as something the future might offer. Analog versions of such were being done by the military for aerial reconaissiance photography. Today's digital versions are a lot smarter.


The interesting thing about film is that even when one reaches some limit on resolution (such as that which a lens might introduce) there is still a certain quality (requiring some sort of new technical name), related to the concepts of resolution and dynamic range, that is transferred when employing more sensor pixels. Or when using more bits per sensor pixel. It's not strictly resolution nor range. But it's certainly a very tangible visible quality. I can see it in 5K transfers of Super8. A visible increase in quality over those transferred at smaller K. Whatever this quality is, it brings the result a lot closer to what the original film looks like when projected.

When one thinks about it this should make sense. How can any information in the film be redundant? The film particles are incredibly small but far more importantly, they are random in size and position, so each and every particle is important, for each must be statistically correlated with the infinitely fine variations in the light. It's not as if the particles jump into some preset position and size with respect to the light (that would allow a finite numerical description), but must encode aspects of the signal (if only in part) that go all the way down to the atomic scale (ultimately requiring some semi-infinite numerical description in order to fully describe). Add to this the frequency at which the images are shown and there is a certain amount of error cancellation going on as well, that both of which, deep within our brains (or wherever consciousness occurs), must connect us (if only in part) with the very being of light and images. So it stands to reason that the more sensor pixels are used in a transfer, the less such deep connections or correlations would be lost. Regardless of film gauge or what such correlations might be technically called.

I think "definition" might be a good word for it. Or perhaps "dynamic definition" since that has no current currency. It doesn't have the strict meaning that "resolution" has acquired, nor that which "dynamic range" has acquired. It can suggest both without becoming strictly one or the other. It is that which is in the complex spatio-temporal relationship between the particles and the frames ... it would be that aspect of the photochemical signal which requires more digital samples be acquired (rather than less), if one is to increase it's visibility in any digital result. Perhaps it's something that must await the invention of quantum computers and related peripherals, in order to properly transfer.

C