Jump to content

Best Super 8 footage you have ever seen (I bet)


Friedemann Wachsmuth
 Share

Recommended Posts

I watched it again and although it's very clean, the colors and latitude look much nicer than digital. You have to think of it in terms of how you would make the stock look, or the other stocks which all have more grain. But it's nice to see how far you can take the format. I would love to shoot a short feature on it for someone if i had the chance, and use all three negatives.

 

Yeah I agree. It doesn't look anything like video. And that theory about sharpening - I can't see any sharpening applied at all. Edges have a nice smooth roll off. No speed bumps at all.

 

The "looks like video" comment was really just a figure of speech - that what was really meant by such is that the work didn't look like Super8 film.

 

And it doesn't.

 

But even when translated, such an observation would be completely meaningless anyway. Who cares if it doesn't look like Super8 film? It doesn't look like the Eiffel Tower either. Nor does it look like my pet dog. Nor my bacon and egg roll. To entertain any of these observations, as some form of meaningful criticism, is completely pointless.

 

C

Link to comment
Share on other sites


What matters in a work is not what it looks like (or what it doesn't look like) but what it is.

 

However what it is, is firstly an image. It is what we see. This is fundamental.

 

And what we see is firstly and fundamentally what is there on the screen - which is not (contrary to some misguided theories) what is in our head. What we see is what the artist has made, rather than what we might make of it.

 

It is only after such that we might add what we might make of it - be it in terms of how it was made (if we know), or how we think it was made (if we don't), or how we interpret the work, or indeed (if we're so inclined) how we might have otherwise made the work (!)

 

But the moment we start assigning more importance to these secondary effects we run the risk of not seeing what is there in the first place - we risk seeing only what is already in our heads. We risk remaining or becoming blind.

 

C

Link to comment
Share on other sites

Regarding use of an unsharp mask, I can see more evidence of that in Villar's work than in Wachsmuth's work.

 

But Villar is otherwise right. One shouldn't overuse unsharp mask (if using it for what it's intended).

 

Indeed these days one shouldn't really use it all. There are far better deconvolution algorithms than the original unsharp mask - typically called "Smart Sharpen". The basis of sharpening algorithms have actually been around for a very long time. They have their origin in analog versions of such. And if used correctly they can make more visible, information in the original data, that was otherwise not as visible. But if used incorrectly (which is far easier to do than otherwise) they will degrade what one has otherwise managed to extract. Pseudo-sharpening. Fake sharp edges. In such cases it is better not to use them at all.

 

But it all depends on what you are after. As an artist you might very well want to degrade the image through use of an unsharp mask (or similar).

 

C.

Link to comment
Share on other sites

Hi Jose -

 

your use of the sharpener is not in any way an abuse of such. And 120/0.8 is by means overuse. Your work is great. In fact it's stunning. Some of the best Super8 I've ever seen.

 

But there's no evidence that Wachsmuth has abused the sharpener either. In fact it looks softer.

 

 

C

Link to comment
Share on other sites

  • Premium Member

The best footage I ever saw in Super8 was about 10 years ago or more from flying spot in seatlle (now Lightpress I think?)

 

Freya

They did amazing work on their Shadow telecine, the only one I knew of modified for Super 8 and HD. Mostly due to their great colorists of course. I flew from Dallas to Seattle once to see them do a transfer for me, good folks. I think they were selling the Shadow a few years ago due to lack of film projects. Must be hard for those shops to invest hundreds of thousands of dollars on a machine and see the resale value go down to $20k.

Link to comment
Share on other sites

Following some research on digital restoration of the Wizard of Oz, I came across something in an article that I've also observed when scanning Super8 - that as you increase the scanning definition, the finer the grain becomes.

 

It's a very counter-intuitive thing but it actually happens. The Oz restoration crew found that they didn't need to employ any grain removal algorithms if they scanned it at 8K. Or to put it another way, scanning at a higher definition was itself a "grain removal" technique.

 

Now the original scan is not itself screened (of course). It's down converted from 8K to a release format (2K or 4K).

 

But it's the round trip that "removes the grain" - scan at 8K, release at 4K (or 2K).

 

But it's not grain removal in the traditional sense. It's not a statistical analysis of the image artificially characterising the image in terms of a signal on the one hand, and noise on the other. Such methods always lead to a softening of image information. Instead it's by extracting more information in the first place - information that is otherwise lost in lower definition scans. Lost in the gap between one pixel and another (so to speak).

 

Its very counter-intuitive but it's the nature of analog images in relation to digital sampling of such. The analog image exhibits less grain as you increase the digital sampling rate.

 

C

Link to comment
Share on other sites

Following some research on digital restoration of the Wizard of Oz, I came across something in an article that I've also observed when scanning Super8 - that as you increase the scanning definition, the finer the grain becomes.

 

It's a very counter-intuitive thing but it actually happens. The Oz restoration crew found that they didn't need to employ any grain removal algorithms if they scanned it at 8K. Or to put it another way, scanning at a higher definition was itself a "grain removal" technique.

 

Now the original scan is not itself screened (of course). It's down converted from 8K to a release format (2K or 4K).

 

But it's the round trip that "removes the grain" - scan at 8K, release at 4K (or 2K).

 

But it's not grain removal in the traditional sense. It's not a statistical analysis of the image artificially characterising the image in terms of a signal on the one hand, and noise on the other. Such methods always lead to a softening of image information. Instead it's by extracting more information in the first place - information that is otherwise lost in lower definition scans. Lost in the gap between one pixel and another (so to speak).

 

Its very counter-intuitive but it's the nature of analog images in relation to digital sampling of such. The analog image exhibits less grain as you increase the digital sampling rate.

 

C

 

 

It's not counter intuitive at all. Higher resolution scans of Super 8 are justified for exactly this reason. As the film becomes more grainy it's actually that much more important to properly resolve that grain so as not to "chunk it up" and alias it. It is true there is not 8K of information in the original source negatives for the Wizard of Oz. That is, as long as you are talking about the image. We need to not think about the image we are scanning but the material itself. The grain itself IS information. To properly scan film you need to properly scan the medium. In this case, the grain.

 

Once you do that, you have a more accurate piece of information which can then be down sampled to your display format.

Link to comment
Share on other sites

  • Premium Member

It also seems that if you apply grain reduction at a higher resolution like 8k (even just a touch) and then scale down to 2k or HD it would be sharper.

 

I do find it interesting that the highest end Hollywood effects shops rarely work in anything higher than 2k (except for some specialty shots). For them it comes down to rendering times and in their tests they don't see a significant gain working higher than 2k even when they outputted to 35mm...or at least a gain worth paying for. Tomorrowland & Ant Man (most Marvel movies) are two movies out now or soon that I know followed a 2k effects workflow. Of course film scans are a totally different animal than computer generated graphics.

 

Of course I love the grain, I just wish compression codecs could handle it better since it seems Vimeo and YouTube are the only ways we can share film projects effectively.

Link to comment
Share on other sites

Yeah, these days effects films like those only have a handful of shots in the movie that aren't vfx shots, and the post-production schedule is very tight, so they really have get the effects done in a surprisingly short amount of time. So they use 2k workflows (and divide the work among many effects houses) for the sake of expediency and cost savings.

Edited by Josh Gladstone
Link to comment
Share on other sites

 

 

It's not counter intuitive at all. Higher resolution scans of Super 8 are justified for exactly this reason. As the film becomes more grainy it's actually that much more important to properly resolve that grain so as not to "chunk it up" and alias it. It is true there is not 8K of information in the original source negatives for the Wizard of Oz. That is, as long as you are talking about the image. We need to not think about the image we are scanning but the material itself. The grain itself IS information. To properly scan film you need to properly scan the medium. In this case, the grain.

 

Once you do that, you have a more accurate piece of information which can then be down sampled to your display format.

 

Yes, I don't know if "counter-intuitive" was the right phrase to use. But I've found many remain resistant to the idea of this phenomena - be it due to their intuition or acquired 'wisdom'.

 

Counter-assumptional?

 

For me this phenomena is not surprising as I've been working with it (and theorising it) for quite a few years now. I've become familiar with it. I can see it. Theorising it is a bit more difficult. Designing methods, techniques, technology to exploit it is difficult. But that's the more interesting side for me.

 

In any case I've encountered a lot of resistance to it. But rather than characterise such resistance as 'wisdom' I've preferred to characterise it as 'intuition', and therefore the phenomena itself (which opposes this wisdom/intuition) as 'counter-intuitive'.

 

Now I must disagree with the idea that there isn't 8K worth of information in Super8. I've scanned Super8 up to 20K (per Super8 width) and there's still information (an image) in the difference between such and a lower definition scan, ie. not just noise in the difference.

 

If intuition says this is not possible then we can say this phenomena (which is entirely real) is counter-intuitive.

 

C

Link to comment
Share on other sites

Now I wouldn't say grain is information.

 

By information we mean a signal, or to put it another way: not noise. And by grain we mean noise.

 

However there is potential for confusion here. For there is also the use of the term "grain" to mean the individual clumps of silver in the emulsion (reduced from silver crystals through development), or indeed by we might even mean individual atoms of silver.

 

However in general, by "grain" we mean that kind of shimmer in the image which we don't otherwise normally see when looking at the world with our own eyeballs. However we can see a form or grain with our eyeballs. In the middle of the night, when we go to use the toilet, and everything is dark, and we haven't turned on the lights, and our eyes have become accustomed to the dark, we can see the world as "grainy". There isn't enough photons around to saturate our retina with a smooth signal. Interestingly enough, despite this grainy world it doesn't look any softer. There's no correlation between grain and sharpness.

 

We refer to this phenomena as "grainy" to distinguish this phenomena from what we normally see in the daytime.

 

But if by "grain" we do in fact mean the individual clumps of silver (or dye clouds) then indeed such is information - which we can characterise in terms of position, size and density/distribution. However there's nothing to say at what scale such information can be deemed resolved by a digital sampling system. Because it's not in any single clump of silver that the information resides. It's in the relationship (or difference) from one point to another that the information resides - that the image resides. And the more pixels you allocate to that difference, the finer becomes that difference. The finer becomes the image.

 

C

Link to comment
Share on other sites

Grain aliasing is an important phenomena.

 

Increasing the definition of a scan, will decrease this grain aliasing.

 

Grain aliasing is a result of interference between the digital sampling rate and the otherwise 'native' grain associated with the source image. The effect of grain aliasing is a magnification of grain (an increase in such) - or more simply: it makes the film (via a transfer) look grainier than it otherwise is. Without this concept (of grain aliasing) one might mistake grain aliasing for grain that is 'native' to the film, and believe that a higher sampling rate couldn't alleviate such - ie. that the grain was hardwired into the film itself.

 

But it's not. It's very easy to mistake grain aliasing for 'native' grain because it's not a signal. It's noise. And as such there's no beats or moire pattern that might give it away as an alias.

 

In many ways, much of the supposed graininess associated with Super8 could very well be a function of this grain aliasing, rather than grain in the film itself. But for anyone who projects Super8 and otherwise compares such to a transfer, they will no doubt get this idea of grain aliasing. They can see with their own eyes that the projected film is less grainier than the transfer.

 

 

C

Link to comment
Share on other sites

It also seems that if you apply grain reduction at a higher resolution like 8k (even just a touch) and then scale down to 2k or HD it would be sharper.

 

I do find it interesting that the highest end Hollywood effects shops rarely work in anything higher than 2k (except for some specialty shots). For them it comes down to rendering times and in their tests they don't see a significant gain working higher than 2k even when they outputted to 35mm...or at least a gain worth paying for. Tomorrowland & Ant Man (most Marvel movies) are two movies out now or soon that I know followed a 2k effects workflow. Of course film scans are a totally different animal than computer generated graphics.

 

Of course I love the grain, I just wish compression codecs could handle it better since it seems Vimeo and YouTube are the only ways we can share film projects effectively.

 

Yes, you can apply grain reduction techniques to the 8K scan prior to down conversion. You'll definitely want to apply something because you can otherwise end up reconstructing the original problem again (ie. grain aliasing).

 

Even just a slight gaussian blur to the 8K scan goes a long way.

 

Traditionally you need a low pass filter applied, prior to down conversion - the idea being that the frequencies above the down conversion frequency don't end up back in the down conversion as a low frequency alias. Down conversion algorithms typically apply some sort of filter anyway - but unless you're coding the algorithm yourself you never quite know what they are doing. In such cases a good eye in conjunction with trial and error on choice of algorithm and/or the parameters of a particular algorithm, provide a solution in such cases - and is often required anyway, even when you do know the details of the algorithms in play, because your eye has access to a brain (your own) containing far better feedback algorithms (millions of years in research and development). The computer algorithms are downright dumb in comparison.

 

The 8K scan (or higher) gives you a kind of virtual space in which to apply all sorts of algorithmic possibilities. And the down-conversion then to erase any side-effects of such work. :)

 

C

Link to comment
Share on other sites

  • 2 weeks later...

Carl Looper,

 

Could you tell me (or us) more about your scanner? How did you achieve 20K???

 

You don't need a 20K camera, if that's what you're thinking.

 

When I say 20K per image width, that doesn't mean I scan the entire frame. I just scan a small area within the frame. The smaller the camera used the smaller the area needs to be.

 

So for example, if using a 5K camera to scan Super8 at 20K per image width, it means scanning a 1.4mm wide rectangle within the Super8 frame.

 

C

Link to comment
Share on other sites

If you have a Super8 projector (or any film projector) capable of sitting on a single frame (without melting the film) you can project the image as you might otherwise do when projecting the film on a wall or screen. You then take a digital camera, without any lens on it, and replace the wall or screen with the camera. The projector is projecting it's image directly onto the camera's sensor where the projected frame area is very much larger than the camera's sensor area.

 

C

Edited by Carl Looper
Link to comment
Share on other sites

 

You don't need a 20K camera, if that's what you're thinking.

 

When I say 20K per image width, that doesn't mean I scan the entire frame. I just scan a small area within the frame. The smaller the camera used the smaller the area needs to be.

 

So for example, if using a 5K camera to scan Super8 at 20K per image width, it means scanning a 1.4mm wide rectangle within the Super8 frame.

 

C

 

So wait a minute, you saying you scanned a small area within a Super 8mm frame at 20K? Using a 5K camera??? I am confused :(

 

I have some Super8 film that I shot last year for my final year university project with the intention to have it scanned at 4K and apply spatial-temporal denoising filtering (accurately to it's grain structure) and compare it to modern digital cameras.

 

My results are still inconclusive but I saw the potential when I got a flat 2K scan from a Spriti Datcine HD scanner.

Link to comment
Share on other sites

 

So wait a minute, you saying you scanned a small area within a Super 8mm frame at 20K? Using a 5K camera??? I am confused :(

 

I have some Super8 film that I shot last year for my final year university project with the intention to have it scanned at 4K and apply spatial-temporal denoising filtering (accurately to it's grain structure) and compare it to modern digital cameras.

 

My results are still inconclusive but I saw the potential when I got a flat 2K scan from a Spriti Datcine HD scanner.

 

The scans I did were just to test the idea of film having some sort of information limit expressible in terms of pixels.

 

To scan Super8 film at 20K per image width, just means scanning the film at a sampling rate of 20K per 5.69 mm. (because Super8 is 5.69mm wide)

 

But I don't have a 20K camera, so how do I test scanning film at this sampling rate?

 

Easy I just don't sample the entire frame. Because I'd need a 20K camera to scan the entire frame. But if I scan a smaller area (eg. 1mm x 1mm) then I can scan that small area suing a 5K camera (5K per mm)

 

And 5K per mm is equal to 28K per 5.69mm, or 28K per Super8 image width. Larger than 20K!

 

After doing the test I compared this scan to one done at a much lower scanning rate. If conventional wisdom dictates that there's no more information in a scan done at a higher rate than a lower one (done at some nominal limit) then the only difference between the two scans would be noise or grain. No image information.

 

But in the difference there was still image information. There was also a lot of noise. But the presence of image information means the lower sampling rate had not exhausted all of the information in the film.

 

It was purely a technical exercise.

 

C

Link to comment
Share on other sites

What DSLR perhaps?

 

Is it possible to see some photos of your set-up? :)

 

Do you think you can do triple HDR scanning??? :D

 

The setup wasn't a dedicated system. It was just an ad hoc one built for the tests and then dismantled.

 

C.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

 Share

×
×
  • Create New...