Jump to content

s8mm hd transfer options


James Meagher

Recommended Posts

  • 2 weeks later...
Flying Spot in Seattle does superb work.

 

www.fsft.com

 

How serendipitous to find this board and thread.

 

I am not a current 8mm hobbyist. But I recently (re)discovered a reel of Super 8mm film that I did for a school project over 30 years ago. I want to digitize it, and dabble with it using AVIsynth, Virtualdub, etc. I also want to be able to extract the best stills possible. To me, all this points to a hi-def, progressive, uncompressed 4:4:4 telecine transfer.

 

I had no idea whether anyone offered this.

 

There are a lot of transfer services, but I am looking for ultra quality, and figure that, for a single 50' reel, I just might indulge myself, if I can find a vendor.

 

I have searched, but, so far, have found only two domestic (USA) candidates: Flying Spot in Seattle (www.fsft.com) and Colorlab in the Washington, D.C. area (www.colorlab.com).

 

I have had some preliminary interaction with both of them, but, haven't heard any comments from a customer about either's progressive, hi-def Super 8 telecine service.

 

Both use very fancy and expensive equipment (Rank, etc.) , and charge hundreds of dollars per hour for their S8 hi-def services. Flying Spot has a "standby" rate, if you don't need a definite, speedy time committment from them. Both are industry-oriented (as opposed to home or hobbyist). I don't say that perjoratively, just informationaly.

 

Just today, I received further communication from Flying Spot, and was disappointed in their message. They told me that they CANNOT provide progressive hi-def digitization at 18 fps. They said they needed to run their equipment at 24 fps to produce progressive scan product with Super 8mm film.

 

Now, I want primarily a "data" product (big avi or mov file), not a video or MPEG2, etc. compressed DVD product; and, therefore, progressive is absolutely a prerequisite.

 

I guess I understand about zero regarding film technology. I really don't know why it matters how fast their machine scans the frames. I have X number of frames on a 3" 50' reel of film, and I want X number of progressive, full color-space (4:4:4) RGB (or YUV), 8 (or 10) bit, uncompressed (or lossless codec) frames delivered as data in an avi or mov file. My (obviously limited) thinking leads me to the conclusion that whether the film is scanned at 18 or 24 (or even 1 or 100) frames per second, I'd still get the same sequence of digitized frames. The bits don't know or care how fast they were produced. Nevertheless, Flying Spot tells me that if I want progressive, they need to run their scanner at 24 fps.

 

I suppose that the speed they run at would effect the header data (metadata), and would embed the frame rate in the file so that, upon playback, the player would know how fast to display the frames. That could explain why I need to care about their (the folks doing the digitization) speed. Perhaps, I could edit the header data to change the (embedded) frame rate information from 24 to 18, and then I would have exactly what I want.

 

Alternatively, ColorLab appears to be much more flexible regarding deliverable products (they will deliver avi or mov-- Flying Spot will deliver only mov-- and are flexible on other factors, such as codec, linear or log, delivery medium, etc.). They are also somewhat more expensive than Flying Spot. I don't know for sure, but think they can produce precisely what I want, without any tweaking from me after I receive the deliverable.

 

Does anyone actually have any direct experience with either of these firms (or any other, for that matter) in digitizing Super 8 to progressive, hi-def digital data?

 

Thank you for any comments.

 

Regards,

 

DG

Link to comment
Share on other sites

  • Site Sponsor

If you get a 24fps progressive transfer you could speed change the 24fps to 18fps with no loss, or ask the fsft or colorlab can run the telecine in data mode which would yield dpx files that you could then set to whatever frame rate you would want.

 

-Rob-

Link to comment
Share on other sites

If you get a 24fps progressive transfer you could speed change the 24fps to 18fps with no loss, or ask the fsft or colorlab can run the telecine in data mode which would yield dpx files that you could then set to whatever frame rate you would want.

 

-Rob-

 

Rob,

 

Thank you for your comments.

 

First, might you say whether Cinelab yet offers the type of hi-def, full color bandwidth S8 xfer I seek? I believe your firm was actually the very first firm I contacted, but you (with integrity and graciousness) suggested that, at that time, Flying Spot FIlm Transfer in Seattle might possibly be more suited to what I am trying to do, as well as more conveniently located to me (since I am in Alaska). I certainly thank you for your previous help, as well as your current comments, and would be pleased to patronize your firm if there is a "fit"..

 

I am not familiar with DPX, but, from a cursory search, I see that DPX is an industry standard file specification for interchange of frame images. It also appears that it is well and openly documented, so I would anticipate no problem for me to be able to use data in that form. In their most recent email, FSFT did, in fact suggest that either DPX or TIFF might be used for deliverable, rather than mov file.

 

If I used DPX, would I expect to receive 3240 individual files (from 3 minutes running time of S8 at 18 fps)? Would TIFF be the same? Is ther any reason to choose one over the other?

 

At the end of the day, the actual film frame data itself would be equivalent in either DPX or TIFF (or even MOV), right? I mean if frame data is produced at a certain resolution, color sampling, and bit depth, the only difference among the delivery alternatives is the packaging (file structure), right?

 

Thank you.

 

Regards,

 

DG

Link to comment
Share on other sites

  • Site Sponsor

We are not yet equipped for HD transfers, unfortunately.

 

 

dpx or tiff files can be the same resolution as a mov file, in general dpx files are RGB not YUV (like mov video files) which is better color fidelity so if fsft can do a data transfer to dpx files that would be the best quality.

 

Furthermore you could get a quicktime file and export a series of tiff or dpx files from Final Cut which you could further manipulate in shake, after effects, etc.

 

-Rob-

Link to comment
Share on other sites

Rob,

 

Thank you for your comments.

 

First, might you say whether Cinelab yet offers the type of hi-def, full color bandwidth S8 xfer I seek? I believe your firm was actually the very first firm I contacted, but you (with integrity and graciousness) suggested that, at that time, Flying Spot FIlm Transfer in Seattle might possibly be more suited to what I am trying to do, as well as more conveniently located to me (since I am in Alaska). I certainly thank you for your previous help, as well as your current comments, and would be pleased to patronize your firm if there is a "fit"..

 

I am not familiar with DPX, but, from a cursory search, I see that DPX is an industry standard file specification for interchange of frame images. It also appears that it is well and openly documented, so I would anticipate no problem for me to be able to use data in that form. In their most recent email, FSFT did, in fact suggest that either DPX or TIFF might be used for deliverable, rather than mov file.

 

If I used DPX, would I expect to receive 3240 individual files (from 3 minutes running time of S8 at 18 fps)? Would TIFF be the same? Is ther any reason to choose one over the other?

 

At the end of the day, the actual film frame data itself would be equivalent in either DPX or TIFF (or even MOV), right? I mean if frame data is produced at a certain resolution, color sampling, and bit depth, the only difference among the delivery alternatives is the packaging (file structure), right?

 

Thank you.

 

Regards,

 

DG

 

 

DG,

 

These questions have complex answers. Transferring to uncompressed HD, either 8 or 10 bit, will require a very fast raid system for editing due to bandwidth requirements, and each bit depth presents its own challenges. Transferring to DPX results in log files that will need conversion to linear color space, in most cases a plug in or addon app to your NLE. Transferring to TIFF might be a good alternative, as most NLEs can assemble a sequence from numbered files.

 

Cheers,

 

Rafael Rivera

Link to comment
Share on other sites

We are not yet equipped for HD transfers, unfortunately.

 

 

dpx or tiff files can be the same resolution as a mov file, in general dpx files are RGB not YUV (like mov video files) which is better color fidelity so if fsft can do a data transfer to dpx files that would be the best quality.

 

Furthermore you could get a quicktime file and export a series of tiff or dpx files from Final Cut which you could further manipulate in shake, after effects, etc.

 

-Rob-

 

 

Rob,

 

isn't there a Super 8 gate on a Spirit in NYC somewhere?? I think at Technicolor.

chris

Link to comment
Share on other sites

  • Site Sponsor
Rob,

 

isn't there a Super 8 gate on a Spirit in NYC somewhere?? I think at Technicolor.

chris

 

 

Yes, I believe Technicolor moves this gate around from facility to facility but it can be scheduled, I think their rate starts around $1800/hr so it is a top shelf option. I think the Shadow (which i feel lags in image quality compared to other HD telecine's for 16mm and 35mm) is a near ideal match for Super-8 HD transfers (I think the shadow is 1440X1080 native imagers) such that a transfer from FSFT will be about as good as you can get out of the medium...

 

-Rob-

Link to comment
Share on other sites

Hello everybody,

 

I am using a 1024x768 machine vision camera for my DIY transfer units.

I capture in uncompressed RGB24.. 8bits.

The camera has a trigger, I can capture at 15fps in full resolution.

Progressive and frame accurate, so capture speed is actualy not important at all.

 

Here's a full res still: E64T, Canon 814 XL-S, made last summer 2007.

http://users.pandora.be/ho-slotcars/Linos/E64T/E64T_007.bmp

 

With some minor post processing it looks like this:

E64T_008.jpg

 

I appreciate your honest opinions....

 

Fred.

Link to comment
Share on other sites

  • Site Sponsor
I appreciate your honest opinions....

 

Fred.

 

 

I think it looks pretty good for a DIY setup but compared to a good HD Cintel or Thompson telecine the pic seems flat and the color is very washed out. The flip side of that is a used Spirit-1 is around $500,000.00 but if you have ever used one you can see why......

 

 

-Rob-

Link to comment
Share on other sites

I think it looks pretty good for a DIY setup but compared to a good HD Cintel or Thompson telecine the pic seems flat and the color is very washed out. The flip side of that is a used Spirit-1 is around $500,000.00 but if you have ever used one you can see why......

 

 

-Rob-

 

Thank you for your opinion!

Are you judging the BMP file or the small one? Or both?

 

I always capture as 'flat' as possible.

A lot can be done in post (curves etc)

 

But of cource I'm just an amateur.

No way I could ever beat a 500,000.00 Spirit of cource. ;)

I have no ambition for that, too.

 

Fred.

Link to comment
Share on other sites

After applying an S-curve and some post sharpening it looks like this:

E64T_curves.jpg

 

Fred.

 

Unfortunately you applied too much of a curve. You've compressed your midtones and lost your highlights and shadow detail is now nil.

 

Tweak with a softer hand, as they say. A little goes a long way.

Link to comment
Share on other sites

  • Premium Member
Hello everybody,

 

I am using a 1024x768 machine vision camera for my DIY transfer units.

I capture in uncompressed RGB24.. 8bits.

The camera has a trigger, I can capture at 15fps in full resolution.

Progressive and frame accurate, so capture speed is actualy not important at all.

 

Here's a full res still: E64T, Canon 814 XL-S, made last summer 2007.

http://users.pandora.be/ho-slotcars/Linos/E64T/E64T_007.bmp

 

With some minor post processing it looks like this:

E64T_008.jpg

 

I appreciate your honest opinions....

 

Fred.

 

The banana lower in the picture has a hot spot on it, aka washed out. The question is, is it that way on the film or did it get lost in transfer? Now as one films a wider and wider shot, it is usually those types of highlights that will blow out more quickly on a lower cost transfer system.

 

If one tries to remedy the situation by lowering the exposure, then the mid tones and darker tones may not pick up as well as they are currently being picked up. So Fred is right in trying to go flat and let the filmmaker decide how to ultimately color correct it. Assuming the end user is capable of such a task.

 

So now the question is, can Fred go just a bit flatter and actually get that highlight to come back, assuming it exists on the film, which it probably does on some microscopic level, which is why film is still around and being used.

 

The higher banana may also have a hot spot on the part that is sloping into the distance but psychologically we let that go because it is going away from us rather than towards us.

Link to comment
Share on other sites

...in general dpx files are RGB not YUV (like mov video files) which is better color fidelity so if fsft can do a data transfer to dpx files that would be the best quality...
IF color sampling is at full bandwidth (4:4:4), THEN aren't RGB and YUV mathematically equivalent? I was thinking that (again, at full color bandwidth) RGB vs. YUV didn't matter with full color bandwidth. But, I welcome being set straight, if I am misunderstanding this stuff.

 

 

...Transferring to DPX results in log files that will need conversion to linear color space, in most cases a plug in or addon app to your NLE. Transferring to TIFF might be a good alternative, as most NLEs can assemble a sequence from numbered files.
For the same bit depth, would I more accurately preserve the dynamic range by using log?

 

I guess it would be easier to work with linear-- if only to save the LU step. I was looking at the Thomson Shadow manual (I found it online), and I think I see in it that the Shadow can produce 16 bit linear RGBK (total of 64 bits per pixel) (by the way, what is "K"? "K" seemed also to be referred to as "Key"). Maybe 16 bit linear is sufficient for dynamic range of S8 film stock from 1975. It probably would be a litle more convenient. The only downside might be disk space. But, I have only 3 minutes of film, and disk space is cheap.

 

 

I think the Shadow (which i feel lags in image quality compared to other HD telecine's for 16mm and 35mm) is a near ideal match for Super-8 HD transfers (I think the shadow is 1440X1080 native imagers) such that a transfer from FSFT will be about as good as you can get out of the medium

Rob, might you expand a little, please? What is it about the Shadow that might "edge out" other more high-end telecines? Is it the resolution or something else? I got the impression that the Shadow used a single line CCD array. I could easily be way off the mark, here.

 

Do you know if the Shadow can capture all the way to the sprocket holes, or does the gate block off some of the frame image?

 

By casual inspection of the manual (found online), I also got the impression that, if one operates in data mode, one could specify the resolution. Thus, I could specify a resolution consistent with the actual aspect ratio of the S8 medium. Perhaps, that is what you were getting at in you remark above. Unless I am totally missing it here.

 

Thank you for any further comments.

 

DG

Link to comment
Share on other sites

The banana lower in the picture has a hot spot on it, aka washed out. The question is, is it that way on the film or did it get lost in transfer?

 

Good feedback! Thank you, Alessandro.

 

It looks like I have showed you all the wrong frame. The hot spot on the banana must be on the film. A tiny bit over exposed. Otherwise I would have seen it on my histogram while capturing, right?

 

Here are a few other shots from that same E64T reel:

E64T_022.jpg

 

E64T_21.jpg

 

E64T_20.jpg

 

I only have an Mpeg4 copy available here and there is a lot of Jpeg compression of cource.

But it gives you an idea.... Personal, I like E64T a lot.. Very nice colors.

 

Fred.

Link to comment
Share on other sites

  • 1 year later...
Are you scanning frame by frame?

 

Hello Steven,

 

Yes, I am scanning frame by frame (machine vision camera with trigger input).

I can capture at 8-10fps in full 1024x768 resolution, on an old HP Pavillion.

 

Please forget the above frames. There was a problem with the monitor from my internet computer.

They are looking horrible on my new monitor. I will replace those frames as soon as possible.

 

Here is the clip:

http://www.vimeo.com/1689844

 

Please download the full Mpeg4 version and watch it with a decent player. Mpeg4 needs some gamma corrections on most Windows systems. Mplayer classic + ffdshow is a great solution for this.

 

Fred.

Link to comment
Share on other sites

Those stills and videos look very good on my Mac Book Pro LCD screen. I have seen online clips of 16mm scanned at "reputable" places with Ranks and Spirits and Cine Glyphs that do not hold a candle to what Fred is doing, INMO.

 

Fred, keep up the good work.

 

S

Link to comment
Share on other sites

Fred, keep up the good work.

 

I certainly will Saul!

 

As an amateur, I am not focussed on theoretical '2K-4K' or 12 bits etc, but on the final result on my computer screen and on my TV. Together with an international group of internet friends, I have developed a good working workflow over the years. I can capture uncompressed at 10fps in full 1024 x 768 resolution -on an old HP Pavillion- and then I run the Avisynth script. The special degraining/sharpening does the magic. :)

 

And of cource I do not show all my mistakes :o

But that's all part of the learning process.

 

Fred.

Edited by Freddy Van de Putte
Link to comment
Share on other sites

  • Site Sponsor

Rob, might you expand a little, please? What is it about the Shadow that might "edge out" other more high-end telecines? Is it the resolution or something else? I got the impression that the Shadow used a single line CCD array. I could easily be way off the mark, here.

 

Do you know if the Shadow can capture all the way to the sprocket holes, or does the gate block off some of the frame image?

 

By casual inspection of the manual (found online), I also got the impression that, if one operates in data mode, one could specify the resolution.

 

 

The Shadow uses a set of line array ccd's just like the Spirit however the imagers are set to 1440 for luma and half for color (rgb) just like the Spirit (which works at 1920 luma and half for each color) I don't think the Shadow edges out the Spirit or Millenium (esp) it just makes nice pic's from S8 and the resolution matches the format pretty well. I don't know about the S8 gate but the Spirit(s) I have operated in the past do not see into the sprocket area on 16mm with a unmodified gate.

 

You can specify the output resolution on a Shadow but anything more than 1440 pixels is interpolated not actual real scan of the film.

 

-Rob-

Link to comment
Share on other sites

  • 1 month later...

Pro8mm offers HD scans in 1080p or 1080i resolution scanned on own new Millenium II HD scanner. We can transfer to uncompressed 10-bit Quicktime files, HD-Cam or Blu Ray. Check out our website at www.pro8mm.com for more info or call at 818-848-5522

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...