|
Author
|
Topic: Review Wolverine Reels2Digital MovieMaker 8mm film digitizer
|
|
|
Dan Esmond
Junior
Posts: 16
From: Lawrence, KS, USA
Registered: Oct 2013
|
posted March 12, 2018 03:06 PM
Hi James,
Mike Spice provided good advice. What cleaning solution did you buy? I'm sure there are several that will work, and as long as it says it's meant for (and safe for) film it should be fine. I used Edwal Anti-stat and had very good results with it. Only caveat I would say with most cleaning solutions is that they're highly volatile so I would do it in a very ventilated area. I did it at the bathroom sink with the fan on--I also kept the lid on the container of fluid except for when I was soaking the cloth.
I cleaned most of my films using an old cotton t-shirt (make sure it's 100% cotton) cut into strips, but you can also buy PEC Pads pretty cheaply on Amazon. Those are meant for cleaning film so you can bet they won't scratch it. The key with either is to keep adjusting it as you go so you're not re-depositing dirt on subsequent sections of film. Might not be a bad idea to keep the cloth saturated with solution while cleaning on a forward pass, then for the rewind use much less so that your cloth can soak up any solution that remains on the film. The Edwal I used evaporates very quickly so I never had that problem.
I think cleaning is really important if you want a good result. Because 8mm film frames are so small, even a single speck of dust can and will show up. You might think "oh well that's just one frame", but the way the Wolverine works, if that speck of dust falls down onto the light table it will show up in lighter scenes of every subsequent frame. This happened to me on several films even though I cleaned them all. People have posted on previous pages some unique methods to prevent this problem (holding the machine up vertically or positioning a brush to gently rub the film right before it enters the gate--wish I had thought of that!
I think a near-essential purchase for anyone using the Wolverine is a set of manual rewinds. You can find them on ebay or just buy an old vintage film editor (you may or may not want to use the editor part of it, but it has rewinds built in). That's what you will use to clean and inspect the reels (more on that in a sec) but as Mike said it's also the best way to rewind each film after it's done transferring. Not only does the wolverine have kind of a wimpy motor, rewinding on it is a PITA because you have to swap the reels around. You could of course use a projector to rewind but it's better to just get a set of rewinds.
I can't comment on the hacked firmware as I never tried it, and I actually finished my transfer project fall of last year before there was much progress on that front. But even if you don't go that route you might email Wolverine support and find out if there are any official firmware updates for your machine. They released two updates during the year last year and their web site doesn't mention anything about them. You have to email them, unfortunately. But they were very responsive every time I sent them an email.
As for the sharpness setting, that's going to come down to personal preference. It defaults to Medium so maybe transfer a small section of film at Medium, High and Low and see what each one looks like to you. For me, the Low setting minimized the noise artifacts and the other settings emphasized them, so Low is what I went with. I tried adding a bit of sharpness back via FCPX but not much. One other note on that (maybe it's been fixed in newer firmware) but my machine would not stay on Low sharpness after a power cycle. Every other setting would stick, but unless I went into the menus and re-selected Low sharpness, the films would end up at Medium... which was so annoying because I wouldn't know until after the film had already transferred. Then I'd have to re-scan it. I ended up putting a Post-It note on my machine to remind me to reselect the sharpness setting.
Another good use of a set of rewinds or an editor is examining the films for sprocket damage, bad splices, etc. The Wolverine absolutely will choke on torn sprocket holes. It will get stuck and just scan the same frame over and over again until you pull the film forward with your fingers. I had several films with such damage and chose to "fix" them using Kodak splice tapes. Those tapes are meant for joining tape edits/cuts together but they have pre-perforated sprocket holes so you can also use them to reinforce torn sprockets. They're meant to be applied to both sides of the film (and you should do that if joining film segments) but for fixing the sprockets enough for the claw in the Wolverine to detect them I found that just applying them to one side worked fine enough.
If your grandpa's films have old tape splices in them it's highly probable they'll have lost their adhesion and will break. Or if they were carelessly applied in the first place the wolverine might choke on them and get stuck. So unless you want to just deal with that by babysitting the machine and moving the film forward when it gets stuck you may want to buy a splicer and a bunch of Kodak PressTapes (easily found on eBay--make sure you get the right type for your film--is it 8mm, Super-8 or a mix of both?)
This is up to you but you may want to explore the Frame Adjust feature in the menu to widen the captured area to include the entire image. I was amazed that my Dad's Kodak Escort 8 camera exposed the film almost all the way to the left edge (between the sprocket holes). It may not seem like a big deal but by including that area in my finished films I am actually able to achieve very close to a 16:9 image, which better fits today's screens. You can always crop it back out later, but I felt it was better to get the entire visible image in the capture.
The only caveat with doing that (especially with Regular 8's huge sprocket holes) is that including them in the capture can throw off the auto-exposure in these machines and darken the image a bit. I usually adjusted the exposure compensation to +1 on my dad's films which I had zoomed out to capture the entire film width. You will just need to experiment with it.
I assume you already know the regular wolverine can't do 8" reels. The new Pro model can, or you can rig something up to do them on the smaller machine, or break the big reels down into several smaller ones.
Well that's probably quite enough for one post. I think it's awesome you're doing this for your grandma--she's going to love it. And you'll probably love it too, even though this kind of project ends up taking a lot of time and quite a bit of money. It's totally worth it though. Hope you'll keep us updated on your progress and feel free to ask any other questions you have.
| IP: Logged
|
|
Werner Ruotsalainen
Film Handler
Posts: 97
From: Helsinki, Finland
Registered: Dec 2017
|
posted March 13, 2018 01:55 PM
quote:
I was amazed that my Dad's Kodak Escort 8 camera exposed the film almost all the way to the left edge (between the sprocket holes). It may not seem like a big deal but by including that area in my finished films I am actually able to achieve very close to a 16:9 image, which better fits today's screens. You can always crop it back out later, but I felt it was better to get the entire visible image in the capture.
The only caveat with doing that (especially with Regular 8's huge sprocket holes) is that including them in the capture can throw off the auto-exposure in these machines and darken the image a bit. I usually adjusted the exposure compensation to +1 on my dad's films which I had zoomed out to capture the entire film width. You will just need to experiment with it.
Yes, several Kodak models did the same; for example, the Brownie. I too made use of the exposed area between the sprocket holes and continuously used +1 exp. compensation. The blacks on the Wolverine easily get "crushed", unlike the highlights, which not that easily burn out (compared to blacks crushing); +0.5EV is, therefore, a standard setting for me. When scanning Brownie material, to account for the global exposure value's decreasing because of the large-area white holes, I dialed in another +0.5EV.
An example of such footage:
my own scan:
16p: https://youtu.be/9z7Cd9A44a0 (direct, non-cropped scanner output with 20->16p framerate header conversion via my ffmpeg script; huge overscan with w=0)
60p ("Optical Flow" in FCX 10.4 + vertical crop): https://youtu.be/at6F70cUihs
Phil Vigeant's scan (it might be a Brownie too; note that he hasn't used as large overscanning as me): https://vimeo.com/238321856/
quote:
My aim is to digitise my Grandpas old 8mm reels (around 20 or so)
If you meant Standard and not Super8: Note that originally Double-8mm film (99% of the currently available Standard 8mm footage) was slit in half, which is inherently a non-precise process. Therefore, your films may get stuck. You will, if this often happens, need to either send in the device to Wolverine so that they widen the film channel or you do it yourself. See my lengthy post on the fourth page of the Somikon topic ( http://8mmforum.film-tech.com/cgi-bin/ubb/ultimatebb.cgi?ubb=get_topic;f=1;t=011449;p=4 ) for more info.
[ March 13, 2018, 05:26 PM: Message edited by: Werner Ruotsalainen ]
| IP: Logged
|
|
|
|
|
|
Laurent Bellier
Junior
Posts: 5
From: Bourg Saint Andéol, France
Registered: Jan 2018
|
posted March 15, 2018 04:51 PM
Hi to all of you,
While understanding we canno't expect more frome this device than what it can delivered, I've however tried to chalange the assumption that the observed noize level is a consequence of the too low bitrate. So I've done exactly the same trial as Werner Ruotsalainen posted February 15, 2018 03:52 PM applied to the same family movie sample that I've poste before:
My first post posted February 13, 2018 04:21 PM was using the Official Somikon Firmware for my somikon scanner (the one issued on October, 17th, 2017, i.e. the Somikon release delivering a 20 frames/s 1440x1080 mp4 movie at 9843 kbits/s)
In the today post, I'm dowloading the modified firmware for the wolverine scanner ( delivering a 30 frames/s 1440x1080 mp4 movie at 16370 kbits/s) (One funny thing: the Somikon device now claim he is a «Wolwerine» scanner when starting)
I've the same conclusion as Werner Ruotsalainen: I cannot really observe that one is better than the other, but I let you making your own opinion.
So I'm wondering if the noize level observed could be also just the illustration of the sensor capability itself? (I mean: Somikon sensor quality looks much much lower than my 10 years old Sony Full HD camcorder)
Based on this trial, I think I will probably come back to the original Somikon Firmware, that will give me smaller files for the same observed quality
Other option: I've seen one another web site (http://letransfert.soforums.com/t1163-mon-t-l-cini-base-de-Raspberry-et-arduino.htm?q=pearl&start=30 ) some recent home made telecine results that seems to give interesting results to me. If you want to have a look, please check on this web site to the post «Posté le: Dim 14 Jan 2018 – 18:56» by «Balzac_40 », an open (an zoom inside) the baby face picture sample: what do you think about it?
(It's just a pity for me that I will probably not find the time for now to try to do that myself.... The scanning time obtaine is presently close to 1 fps, so two time less than with somikon/wolverine scanners)
--------------------
Laurent
| IP: Logged
|
|
|
|
Laurent Bellier
Junior
Posts: 5
From: Bourg Saint Andéol, France
Registered: Jan 2018
|
posted March 22, 2018 05:16 PM
Yes, sorry, you're right, I'm trying to get in touch with the author to understand if it's a crop part of original picture, or a downsized version ....
Meanwhile, I've found another exemple which make me dreeeming about (The guy here seems to be really skilled in several areas optic, mechanic, electronic, sofware ....):
- Amazing still picture 1388 × 1036 extracted from a super 8 movie: http://www.cine2digits.co.uk/wed1.png
- video sample here: https://vimeo.com/49963017 and here: www.vine.me.uk/downloads/florida01.avi
(All tehcnical details here: http://www.cine2digits.co.uk/)
Well, I'm now wondering if I can find some time for trying building an home made machine, it seems such good quality can be achieved ....
--------------------
Laurent
| IP: Logged
|
|
|
|
|
|
|
|
|
|
Werner Ruotsalainen
Film Handler
Posts: 97
From: Helsinki, Finland
Registered: Dec 2017
|
posted March 24, 2018 10:06 AM
Folks, I've done some VERY important experimenting with my Somikon now that I've run into a VERY contrasty, B/W scene (a road sign) in one of the films I've digitized. (The original film is available in its entirety at https://youtu.be/SIhQNZcyaXg (60p processed with FCPX 10.4) / https://youtu.be/UQwxYlMbLHE (18p original). Note that I also provide a link to an archive containing all the test scans & framegrabs below - in order to review the results, you will want to prefer that one to the entire film.) This problem MAY affect Wolverine-branded (non-European) units too (or, for that matter, Reflecta ones) - I don't know.
Basically, I wanted to find out whether I can reduce the JPEG compression artefacts of the contrasty sign with the hacked 1440x1080 16 Mbps firmware ( http://retromania.pandelground.com/wp-content/uploads/2017/12/FWDV180N_1440x108 0_24fps_8000.zip linked from http://retromania.pandelground.com/wolverine-reels2digital-moviemaker) and how the sharpness settings affect the image quality.
During this, I've made a very important discovery: the now-latest (Oct/17; NX4294_12_163534) official firmware does NOT support sharpness changing at all. ALL the three settings result in exactly the same output! You can freely check this out in the following archive file:
https://drive.google.com/open?id=1qSoS5wrfKWEPurYfARtJduaxXGlahu7z
To see the scans I've made with the official firmware, look into the two subdirectories of "FW sharpness compare/ORIGINAL". They have three (in subdirectory "1") / two times two (in "2 power cycling between altering the setting") samples: the original (unaltered) scans themselves and the framegrab of exactly the same frame from these videos as .png files while playing back in QuickTume. If you look at the latter, you can see there is basically no difference between them, regardless of the sharpness settings I've used. (The filenames also contain the value of the setting.) The (very similar) file sizes also prove this. (The sharper (sharpened) the image, the bigger the filesize.)
Now, compare all this to the contents of the two subdirectories of "FW sharpness compare/HACKED", which has the "hacked" scans. Between these tests, I've even re-flashed the original firmware to be absolutely sure there is a difference in how the "Sharpness" setting is handled, as is also explained in the directory name "2. test after a full re-flashing (installing the original FW for testing and, then, again tha hacked)". Also note that the second subdirectory has separate subdirectories for English and German as the GUI language. (My scanner was, at first, set to German; I then re-set it to English after a bunch of tests to see whether the bug is language-dependent but, say, the English version is properly implemented. (It isn't.)).
As you can see, there is some major difference between the hacked version's files. This is pretty perfectly visible from the framegrabs' filesize too: 6, 7.2 and 8.1 Mbytes for the low / normal / high sharpness settings, respectively. That is, the low-sharpness version indeed has a lot less (false) information as the sharper ones.
The explanation for the (earlier) hacked version to work, unlike the (later-released) official one? Easy: the latter firmware wasn't properly debugged before release. The developers simply didn't notice Sharpness isn't at all working any more. This problem is very common in the world of programming.
All in all: at least if you have a Somikon with the latest (20 fps) firmware and do want to disable all kinds of in-scanner sharpening (that is, want to use the Low Sharpness setting, as recommended), do check whether the different sharpness settings indeed result in visibly different in-scanner sharpening. If you have the same results as me, just install the hacked firmware.
[ March 24, 2018, 01:20 PM: Message edited by: Werner Ruotsalainen ]
| IP: Logged
|
|
|
|
|
|
Werner Ruotsalainen
Film Handler
Posts: 97
From: Helsinki, Finland
Registered: Dec 2017
|
posted March 25, 2018 07:58 AM
BTW, let me elaborate on another slight advantage of using the hacked firmware caused by FCPX’s not allowing one to process 8mm films with their native speed (16/18 frames per second). Up until now, in order to avoid creating intermediate frames, we needed to re-pack (essentially, change the fps header – nothing else!) the 20 fps output of the scanner into a (bogus) 24 (or 30 etc. - anything natively supported by FCPX) video file and import it instead. (Again, with my ffmpeg scripts above, it’s very simple; however, it’s still a separate step we ALWAYS needed to do.)
Now that the hacked firmware’s output’s framerate is 30fps, which (unlike the 20fps of the latest official and Sharpness-buggy firmware) is natively supported by FCPX (and all the other video editor apps), we can directly import these files into FCPX: no intermediate conversion (re-flagging to 24(...anything supported) fps) needs to be done. This is certainly an advantage.
Of course, if you do want to export the file in native 16/18 fps, it isn’t possible in FCPX.
Actually, I only know of DaVinci Resolve (as of v12.4+) to support these two standard 8mm framerates – see https://vimeo.com/248538512/description . And, AFAIK, even it doesn’t support 12fps, which some R8 shooters did use back then (for example, the soldier that made these, for army admirers/devotees, pretty interesting shots in 1964: https://youtu.be/xKidGxNgn14 / https://youtu.be/ntez1LH-bYE (12p)) to save some film.
You need to slow the video down (to 60% for S8 and 53,33% for R8) from inside FCPX. Unfortunately, my ffmeg scripts will NOT work for FCPX’s output; that is, you can’t just export the file with the native 30 fps and just use ffmpeg to re-flag it to 16/18 fps. If you try doing this, there will be some heavy stuttering in the stream: ffmpeg will swap previous and next frames, making the output useless. This is a major problem.
This is why you must do the slowdown from inside FCPX. This, however, will still physically export a 30 fps file (or the framerate you’ve explicitly chosen when creating the project), meaning slowdown will introduce doubled frames to be displayed (A.K.A. pull down). You can always change the default “Normal” “Video Quality” setting (the one just inserting “dumb” doubled frames) to, say, “Optical Flow”. (This is what I personally do for all my FCPX-processed files. Plus, I create the project at 60p to allow for as fluid movement as possible. After all, if, because of FCPX’s limitations, I do need to insert computed intraframes in between original frames, I try to make use of this to make the movement much-much fluider. (This is especially important with 12fps footage. The pans (or, for that matter, any camera movement) in the two 12fps army films I’ve linked to above look absolutely horrible at 12fps. At 60fps, apart from the artefacts, it’s much-much more eye-friendly, as you can also check out at https://www.youtube.com/watch?v=h3XBthKlNPs / https://www.youtube.com/watch?v=rBGSCtX4xo4 .)
This is why I always upload Optical Flow-processed 60p 8mm scans in addition to the direct scanner output (12/16/18fps) on my YouTube stream: https://www.youtube.com/channel/UCSJ3JqZMyf0LDLNofpBtzow/videos .
Speaking of frame interpolation problems, the “Frame rate conversion” at http://www.thebattles.net/video/8mm_restoration.html also explains the problems you’ll face when you use “Optical Flow”.
| IP: Logged
|
|
|
|
Werner Ruotsalainen
Film Handler
Posts: 97
From: Helsinki, Finland
Registered: Dec 2017
|
posted March 29, 2018 02:29 PM
quote:
I currently have a troublesome reel that I'm going to scan backwards. Maybe the claw will get a better grip on the opposite side of the sprocket holes.
Yup, I also do the same, particularly with spools that, otherwise, I couldn't scan without wobbling in 5-6 attempts. Then, generally, a reverse scan saves the day.
quote:
Is there a modification for this fps changing Mac script that can also reverse the MP4 at the same time?
There surely is, but, unlike my simple fps changer script, it also reencodes the video (takes a LOT more time & decreases quality):
ffmpeg -i inputfile.mp4 -vf reverse reversed.mp4
Therefore, I wouldn't use it as a separate step if I planned to another step of post processing; for example, vertical flipping. (At least in QT under MacOS, flipping is very CPU-intensive.) I can post my FCPX workflow for reversed scans if interested.
| IP: Logged
|
|
|
|
|
|
Werner Ruotsalainen
Film Handler
Posts: 97
From: Helsinki, Finland
Registered: Dec 2017
|
posted March 30, 2018 06:59 PM
quote:
I have to say the difference is dramatic in terms of frame stability in the scan...
Yup, the same is my experience - and also some of Amazon users. See my previous posts / photos on exactly this matter.
quote:
Is there something I can use to keep the higher bitrate or am I just confusing myself?
While i haven't tested the app, I assume it's just re-muxing the video files (just like my ffmpeg scripts) but does not recompress them. Then, the figures you posted are perfectly normal: let's not forget that the higher figure is for 30 fps, while the lower one is for 15/18.
For example, the just-uploaded 16p video at https://youtu.be/ID9aRrAOrdY is 8,74 Mbps after the simple 30->16 header conversion (via my ffmpeg script), while the 30fps original is 16,38 Mbps. The ratio between the two bitrates is exactly that of the framerates.
All in all, don't be afraid.
(Again, this assumes the app really doesn't reencode anything.)
| IP: Logged
|
|
|
|
|
|
|
|
|
|
|
UBBFriend: Email this page to someone!
Printer-friendly view of this topic
