Alternative to convert Super8, 16mm and 9.5mm films to digital video using inexpensive flatbed photo scanner and frame extraction free software. Now with optical sound extraction. There is also a new free software for 35mm films under development with a beta version.
"As well as the seeds are planted in the earth and sprinkled with water to grow trees that give fruits to feed us, the silver seeds are planted in the film and sprinkled with light to grow on the screen forming images that tell stories to feed us too"
1 - The softwares are free
2 - The Flatbed Photo Scanner is inexpensive
3 - Great results scanning negative or positive films due to the scanner is improved to both.
4 - This method can keep the beauty of film grain instead of bluring it.
5 - Do not need time consuming and difficult projector modifications
6 - The scanner can do HDR scan and manual exposure with levels control
7 - Scanner is useful for other tasks like photo negative, slide and print scans, a good investiment
8 - The process can be automated if you build a film transport hardware
9 - The scanner is portable and works with notebooks
10 - The resolution is good for 720p or 1080p HD output.
11 - The software works on average computers
12 - The software has options for all aspect ratios, R8, S8, 9.5mm, 16mm, S16, U16, and also "All The Gate"
13 - The film guide to run the film through the scanner is simple and easy to build
14 - The scanner software has manual or automatic setttings for scan, so you can do image correction in post production or in the scanning moment making things easier and faster if you want.
15 - You can do other tasks in the same computer while you scan the film, you just need to run the film, click scan, wait scan, and do it again, file naming and file saving are automatic.
16 - This method keeps the perfect 24p frame cadence, something difficult to get doing realtime telecine with an unmodified or even modified projector. This method is the same thing of a frame by frame filmscan, a thing serious filmmakers dream about. You can get this with a projector, but you will need a frame by frame projector, or a very slow motion projector and rebuild the movie in video editing software.
17 - the scanner and software can extract all the image in the film frames, without cropping it
18 - new feature added: optical sound extraction
(the numbers presented here may vary depending on the scanner used)
If you work with a big transparency lid scanner like the Epson V700 and V750 and low dpi setting, like 3200 dpi, you can scan one 50 feet super 8 cartridge in one hour work, and one 100 feet 16mm reel in two hour work, good for a DVD quality output or to share on internet.
After some new tests i found 3200 dpi is the best quality/performance resolution for DVD and WEB. Also using unsharp mask in scanner software gives good quality and faster workflow to improve the image. You can chose low, medium or heavy unsharp mask in EpsonScan software.
But, if you want Full HD quality (Bluray output)...
Working at 9600 dpi for Super 8 or 6400 dpi for 16mm film, the highest quality possible, It will take 1 minute and 50 seconds do to each scan, counting the time to run the film manually, click scan and the scanner work itself, and the number of frames per scan will reduce signifcantly due to limits in file size, increasing the number of scans. So it will be about 8 hours work to scan one 50 feet Super 8 cartridge or 16 hours work to scan one 100 feet 16mm reel, if you do not stop. DIY work means you really need to work, so if you think you can do it, go on and read the full text bellow, all you need to know is here. (remember you can build the hardware transport to leave the system scanning all the reel alone, unsupervised). Then take a deep breath and go ahead...
When I was researching about negative / slide scanners I realized it could be used to scan Super8 and 16mm films, because the scan quality was good and a flatbed scanner could scan many frames at a time.
But I realize the challenge would be how to extract the frames from the scanned filmstrips and how to build a movie with good stabilization and in a good / fast workflow?
So I started google about super8 telecine / scan / scanning.
I found some unsuccessful try outs but suddenly I was pointed to a German super8 enthusiast who developed a great solution to extract frames from scanned filmstrips with great quality and stabilization.
I downloaded his free software and did some trys with the provided film strips he included in the software. The software was very mature in development, version 5, with the main features ready to do the job for regular 8 and super 8. The idea is simple and smart, the software uses the sprocket holes as reference to find, align and extract the frames.
I became excited about the results I got and started talking to him by email about it, and I asked him to include an option to extract Max8 (Super Duper 8) and to correct some small bugs I found.
He answered kindly and gentle and started working hard to improve the software.
So after this first improvements I e-mailed him with the idea: "16mm extraction - freedom for indie filmmakers" and he got excited and started developing it
After some weeks he came up with an early version, now a .net framework instead of java software. I do not know anything about computer programming, but I know how to test it carefully, and wich features to request to cine production. This was the way we worked togheter.
We started an intensive e-mail exchange, about 300 e-mails since the beginning, many versions have been developed for me to test to find bugs and to give him ideas of aspect ratios and features.
When I saw a perfect functional version working I decided to get a film scanner. So I bought the flatbed Epson V500, the cheaper one, after reading lots of reviews.
He finished the bug corrections and features implementation and merged Super8 and 16mm extraction into one software and then Wolfgang Kurz called it CineToVidPro 1.0 version. The work with high dpi resolution became more stable also. Now the version 1.2 also works with Pathe 9.5mm films and a new software for 35mm films are under developing with a beta version available.
You can get the free software and talk to him at:
Before contacting Kurz, read the help files and also the tutorials and sites. Also try the software to make it works. In this way you will have a better chance / opportunity to improve yourself.
In his websites you can find the download links and also the explanations how the software works (detailed help files). Also demo videos and, if you do not have a scanner or a film to scan, there are scanned filmstrips to download and do tests.
To transfer your films to video, all you need is a flatbed photo / film (transparency) scanner pluged to your windows computer, and CineToVidPro (also the .net framework). I recommend 4GB RAM for high dpi work.
Build a guide to film runs through the scanner glass. I did the filmguide with two layers of black adesive tape glued to a piece of glass and put this piece of glass over the scanner glass, and film runs between this glass sandwich on top and bottom and between the two black adesive tapes at sides.
You can see the film guide I built in the photos here in this page. It is so simple, and you can improve it and also use film rolls to feed the scanner. The important point is to built it very well aligned to scanner edges to get scanned strip straight, not angled. You may also find a way to avoid dust particles enter between the glasses. Clean your workplace, scanner and film before scan work, and while you feed the film, keep cleaning it. You can see some parts in this video is cleaner because I did a better cleaning job.
To avoid light changing from one strip to other I used manual exposure in the EpsonScan software, and I keeped all settings flat. I keeped the same locked manual exposure for all the film reel.
You can improve the exposure to get better dynamic range keeping good shadows and highliths with this tip: Select auto exposure in scanner software. In scanner preview window select the filmstrip without the sprocket holes, because the light trough sprocket holes ruins the auto exposure. Open the levels window. Copy the red, green and blue values. Go back to preview window. Select the filmstrip with the sprocket holes. Selec manual exposure in scanner software. Open the levels window again and type the values found before. This manual exposure tip also works great for color correction, because the levels values found by auto exposure corrects wrong colors.
Count how many frames your scanner can get at each scan. The V500 can scan 16 frames per scan in Super8, 14 frames useful for extraction, so after each scan I run the film 14 frames and scan again. (The V700 and V750 can scan about 47 Super8 frames in each scan).
If you do not want to run the film manually, you can build an automatic film transport hardware controlled by a free software. The tutorial and software are in these links:
After scanning you go to cinetovidpro software and extract the frames from the scanned filmstrips. Cinetovidpro can also generate a movie, or you can import the frames to your video editing software.
Tips for advanced users:
Remember this: 3200 dpi is the best quality/performance resolution for DVD and WEB. Also using unsharp mask in scanner software gives good quality and faster workflow to improve the image. You can chose low, medium or heavy unsharp mask in EpsonScan software.
If you are photo/video editing expert and want the best possible image the following tips are useful:
The more dpi you scan the better the quality you will get, because the scans needs lots of unsharp mask filter, and unsharp mask works better with more dpi, more pixels. Doing a 50% downsize in resolution (half width and half height in image size) after apply heavy USM also helps a lot to avoid artifacting. After the downsize you can resize for the resolution you want. The unsharp mask, downsize and resize must be done in photo or video editing software after Cinetovidpro extracts the frames. In photo editing software you can use a batch processing to do it in all still images automaticaly. You can also clear chroma noise with a filter, but take care about cleaning luminance noise because it can hurts the beauty of the film grain, do it carefully if needed. There is a limit in the resolution size I found: the scanned strip height x width needs to be less than 134 megapixels. Working with the V500 scanner, 9600 dpi is the best choice for super8 and 6400 dpi is the best choice for 16mm, considering scanning horizontal direction, 68mm x 9mm for super8 and 68mm x 17mm for 16mm.
You may ask: why not scan in vertical direction to get more frames at a time? The answer: Because the vertical scanning takes loooonnnnnggggg time and also there is a problem, the film would need to cross all the scanner vertically and it would cover the scanner calibration area and the machine would stop working. It does not deserves the try. Also, working with high dpi (9600 and 6400) will not let you scan more than the V500 horizontal area due to the file size limits the software have. But, if you want to work with low dpi (3200) and output DVD or WEB quality, you can use more area and scan more frames in each scan using scanners with bigger horizontal area, like the V700 and V750, this scanners are a little expensive, but wotking with 3200 dpi it will make things much, much faster for you, reducing about a 8 (eight) hour work to about a 1 (one) hour work. Also these scanners have a sharper image.
It seems the lowest dpi resolution for CineToVidPro software is 3200 dpi, to work with 2400 dpi you must use the CineToVid version 5 Java software. I recomend to use the Cinetovidpro.
To see a 2400 dpi scanning results go to vimeo.com/28249965
I also included here, in the photos, some crops from the scanned super8 frame showing the image tweaking process, step by step after scanning. The settings used for it was: Unsharp Mask: 400 / 4 / 0 (this is for 9600 dpi, for other dpi USM will be different, the less dpi, the less unsharp mask you can do). I did a test with photo negative films and I found for 6400 dpi the unsharp mask settings 400 / 2 / 0 is great, so this can be the settings for 16mm films scanned at 6400 dpi. These settings are for the V500 scanner, for other scanners the settings can be different.
Why to do a 50% downsize? To clean the artifacts generated by unsharp mask, compare the crops and you will understand. (click in the images showed in this page to open it on flickr and see bigger)
If you want to see the tweaking process step by step go to this flickr photo album:
If you think this unsharp mask / image tweaking method is difficult or time consuming, you can just apply low, medium or heavy unsharp mask in the EpsonScan software when scanning.
I found the Kodachrome reacts better to scanning, the scanned dynamic range is great. The Ektachrome, when I get good highlights I loose details in shadows, and when I get good shadows I blow out the highlights. Here, in this video, I decided to preserve the high lights due to sea waves. I am developing a solution for this. The idea is to do 3 scannings per strip, in each position I put the film I can do 3 scans, optimized for shadow, midtones and highlights. So I can merge the scans in a free HDR software before frame extraction. In this uploaded test you can see the grain from Ektachrome is wonderful (to people who likes film grain) and deserves the tweaking. In the EpsonScan software there is a way to save exposure settings and select it from a list of saved ones.
The HDR scanning takes much more time to do and I found doing a carefull adjust in the LEVELS in the EpsonScan software in Manual Mode can give a great average result. Use the manual levels adjust described above and it will be fine. (Read the exposure tip in the beginning of this tutorial)
No secrets here, I used VirtualDub and DeShaker plugin, both free software, after video generated, to correct amateur shaking camera operation, these movies are from common people in family tours. CineToVidPro has an option to scan a larger area to be croped when doing DeShaking, so you do not lose too much area from the original image after this. If you think the amateur shaking are charming, keep it. The DeShaker plugin can also be used to stabilize just the gate, not the image itself, using the sprocket hole as reference. Links for the softwares:
The movies scanned in this test was 18fps and it was edited at 24fps. Due to this the motion is a little fast speed. To avoid this the CineToVidPro software also have options to convert 18fps Super8 to 25fps or to 24fps and convert 24fps to 25fps. It does this repeating some extracted frames, called fill frames. Using DeShaker you can automatically reposition the repeated (fill) frames and recriate the perfect frame cadence. To do this you can follow some instructions from Kurz: In the second path of DeShaker use pretty high values for smoothing pan, rotation and zooming (4000 or even above).
To learn how to use CineToVidPro just read the help file included in the download and also the websites pointed here. Help files are in english and german, very well explained by Wolfgang Kurz. If you do not have a film to scan or a scanner you can try the software using the filmstrips he provides for download in his website.
There is a group on vimeo to share films telecined / scanned with cinetovidpro and to talk about the software workflow. Be welcome to join:
I invited Wolfgang Kurz to sign in to Vimeo and now he is a member of the CineToVidPro group. It is great to have the software creator and developer among us.
I also dedicate this first video to Kurz for his great work creating and developing the software, for the colaboration we have each other, and for the polite relationship we use to be.
Any comment or feedback feel free to share, and also, if you use the software, put the tag cinetovidpro in the uploaded video and add it to the group.
Hope this can help the Super8 and 16mm community.