In a nutshetll this guy constructed a lil device taht moves this film feeder mechanism. What it does is tells the computer to initate a scan, once the scan finishes it moves the film and once it gets a new section on there pow it scans agian til it's finished take a look. It's wild stuff.
I looked into the whole scanner idea long ago but the results I was getting were pretty much on par with what he shows in his sample AVI. The problem is that the numbers regarding resolution of the scanner don't really tell the whole story. When you crop into an 8mm section of even a standard slide shot with a Nikon, the end result is pretty dismal, even with the best commercial scanner out there (I've tried). I think the idea is fun but, in practicality, the final image just doesn't justify the time. Softness is something that will always be a problem until scanners end up with a much higher effective resolution that is asthetically pleasing as opposed to only meeting basic mathmatical criteria that doesn't take into account how the image really looks. I have looked into making a lens array that would bring the image size up to something like 4x5 so that a cheap scanner could get better results but, in the end, I'm not convinced the life of a standard scanner would equate to a single roll of super 8 film. That's over 3600 scans! I would imagine that the unit would go through a certain amount of wear and change from one the beginning of the roll to the end. Could be wrong but, really, who wants to spend a week digiting a single 50 foot roll of film? 8O
In my opinion a Super-8 Scanner would have to have 12 times the resolution than a 35mm Scanner because the area that is being scanned is 12 times smaller.
So a high quality 35mm Scanner is actually a low quality Super-8 Scanner.
Roger, does it matter which way the film strip lays on the scanner? If the film strip is turned 90 degrees, does the quality of the scan change in any way?
I did some scans of 9.5mm film frames with a 2400 dpi flatbed scanner.
9.5 frame sizes are 6.5 x 8.5 mm. at 2400 dpi, the capured frame have a resolution next to 765 x 553 pixels (nearly PAL resolution) and the quality is really good.
So, to scan a S8 frame the minimun resolution of the scanner must to be 3600 dpi.
a good 35m scanner is a good s8 scanner. There is no more information in a s8 frame than in a crop of a 35mm slide..
At roughly 4000 dpi you get almost everything out of film. Until that point, higher scanning resolution results in significantly better scans. If ypu go from 4000 to 8000 dpi, the difference in extracted detail is tiny, especially compared to the HUGE filesizes. A drum scan sucks out evrything, but is very expensive, results in incredibly huge file sizes, and most of the time not an option due to the effort.
Much more important is dynamic range and color depth. Scanners have 48bit..
In a word: at 4000 dpi you will have 95% of the information, the remaining 5% are a PITA to get out of the film.
jean wrote:a good 35m scanner is a good s8 scanner.
I disagree. A good 35mm scanner will scan super 8 but whether it will provide "good" results has yet to ever be proven. I have seen people debate this idea endlessly on all kinds of forums but the fact is that the end results when scanning super 8 with a flat bed are disappointing, at least to me.
jean wrote:There is no more information in a s8 frame than in a crop of a 35mm slide.
I agree. Which is why an 8mm crop of a 35mm frame looks just as soft as scanning the entire 8mm frame with the same flatbed scanner. Again, what the numbers say is theoretically possible and what the results are asthetically seem to be totally different, as far as I can tell. I would love to see results that prove differently but, so far, all I ever see are people discussing numbers and never a single decent image that backs up the concept.
compaired to spending a grand + for a workprinter (+ the dv camera) than spending a few hundred on a scanner and external hard drive, i'd go with the scanner.
camera8mm wrote:compaired to spending a grand for a workprinter than spending a few hundred on a scanner and external hard drive, i'd go with the scanner.
Isn't sharpness an issue? I mean, the flatbed may be cheaper but not getting what you need at half the price is hardly a bargain, I think. ;)
MovieStuff wrote:
Isn't sharpness an issue? I mean, the flatbed may be cheaper but not getting what you need at half the price is hardly a bargain, I think. ;)
Plus the film scanner will burn out quickly and in the end a workprinter will be much cheaper since it will last.
jean wrote:a good 35m scanner is a good s8 scanner.
I disagree. A good 35mm scanner will scan super 8 but whether it will provide "good" results has yet to ever be proven.
Of course I meant this in an "academic" way, referring to resolution only. Unfortunately no 35mm film scanner is suited for s8 transfer.
Flatbed scanners (I don't know about the latest 4800 epson) are no match for dedicated filmscanners, although they are the only ones where one can put some s8 through without beeing an engineer. The resolution numbers especially for flatbeds are purely marketing hype, when I remember correctly the epson 3200 flatbed has a real resolution of some 1600 dpi, which is not very much. I can post some stills scanned at 3600 dpi, if anyone cares to see them..
Echoing Roger's statement that a good 35mm scanner does not necessarily equate a good S8 scanner, and working from 1600dpi, we can deduce that:
35mm, at 24mmx36mm (for a 35mm STILL neg) would yield a 1477x2215 image. (1/26x24x1600)
35mm, at 24mmx18mm (for a 35mm MP neg) would yield a 1477x1108 image.
S8, being approx 1/16 the size of a 35mm MP neg, would yield 370x277...which is about 1/4 the res of what a workprinter would do with a standard NTSC camera at 720x480.
To get the resolution you are looking for, you would need a scan resolution of 6400dpi.
Brandt wrote:
S8, being approx 1/16 the size of a 35mm MP neg, would yield 370x277...which is about 1/4 the res of what a workprinter would do with a standard NTSC camera at 720x480.
I've never done the math because it all seemed sort of academic to me. I'm for what ever looks the best, regardless of what the numbers state. But if what you say is true (370x277) then that sort of explains why any super 8 scans I see from a flatbed look like a frame grab from VHS.