0

I'm looking to digitize a couple of Super 8 film reels using what's called a Wolverine digitizer/converter, which is a device that allows a person to digitize Super 8 film. However, while researching what the reviews had to say, I stumbled upon this paragraph:

[In regard to the image posted by user Laurent Bellier on 8mm Forum regarding the Wolverine device] Neat Video and other software plugins can help lessen the prominence of the compression artifacts. But they can never recover the detail lost to those artifacts in the first place. It's a garbage in->garbage out type situation. The problem with these machines is they produce mp4 files that are bitrate-starved. Compression artifacts are the result when there is too much detail for the capture settings to reproduce accurately. Yes you can make them look a little bit better but that does not fix the actual underlying problem: that detail was never captured by the wolverine in the first place--it's permanently lost.

I'm already somewhat familiar with the concept of bitrate. Bitrate, if I understand correctly, is the amount of information from a video that is being shown on the screen per second. That is why it is generally better to watch a movie on Blu-ray rather than on Netflix because Netflix uses compression algorithms that are much more aggressive since internet speeds are a limiting factors. By watching the Blu-ray version, you are experiencing the movie at a higher "fidelity," since it's being shown at a higher bitrate. In other words, the Blu-ray offers a "truer" version of the movie compared to the Netflix version.

But what I don't understand is what bitrate has to do with MP4 compression artifacts.

Also, what is meant by bitrate in the above case?

The way the above paragraph uses the word "bitrate," isn't bitrate the amount of information that is being captured per second from each frame before the video file is created before applying MP4 compression?

If my understanding of the way the word bitrate is used in the above paragraph is correct, what does the person mean when the person says that the Wolverine device creates MP4 files that are "bitrate-starved"? Does the phrase "bitrate-starved" mean that the device is recording each frame at a low bitrate?

If bitrate-starved means that the device is recording each frame at a low bitrate, what is the cause of this? Is a low bitrate caused by the machine scanning each frame too quickly, resulting in a low bitrate? Is a low bitrate caused by the light source not being bright enough to capture all of the information in each frame resulting in a low bitrate?

And, finally, what do MP4 compression artifacts have to do with a low bitrate of an uncompressed video file? Does it have to do with there being too little information (low bitrate) in each frame? Is that what the person means by "bitrate-starved"?

I really need help with this. I need to know the pros/cons of buying a consumer device vs. going with a professional service. The consumer device is cheaper but riskier as it uses the sprockets to move the film frame by frame, which can further damage old film.

3
  • There’s also the issue of colour depth. A lot of consumer devices only use 8 bits of data to define the colour of each pixel. Some use less data to define red and blue pixels compared to green pixels as a way of reducing the amount of information that needs to be stored. This also reduces the ability to do image correction afterwards, regardless of overall bandwidth of the file.
    – tomh
    Sep 4 at 6:27
  • Yep, I'm already familiar with the concept of color depth. I have a scanner that I plan to use to scan a bunch of old family photographs. Assuming the device does not use less data for the red or blue pixels, 8 bits means 2^8 possible colors per channel. Since there are 3 channels that's 2^8*2^8*2^8 colors. But what I don't understand is that even if I could find a device that stored more than 8 bits per channel, wouldn't there be no way of even representing those added colors on a computer monitor? So, if your computer monitor can't show these colors, what's the point of scanning >8 bits? Sep 4 at 8:44
  • Because the image’s colour is defined with greater precision. So you as the grader get to decide what information is shown on screen and what is discarded when you grade your image. Eg you can pull more detail out of darker shadows if you need to make the image brighter, even though you couldn’t see that detail in the original ungraded image on your screen.
    – tomh
    Sep 4 at 8:48
1

I don't own this scanner, but looking at the specs, it appears to record in MJPEG mode, scanning one frame at a time.

The MJPEG codec records each frame of video as a JPEG image. So each frame is compressed using JPEG, which is a lossy compression process.

It then (I think?) has some software which can take the large MJPEG files and compress them further using mp4 h264? compression.

MJPEG is a bit like a flick book of JPEG images - each one is compressed individually - there is no compression across multiple frames. This means it takes up more space than mpeg-4 compression which does analyse across frames, and so can compress videos further.

With all lossy compression systems, the more detail there is an image, the harder it is to compress it efficiently. So if an image has lots of speckles and noise on it (like old film often does), this will either make the file much larger, or if the compressor is trying to target a given file size, the areas of low detail will become blocky.

What doesn't tend to happen with video compression is added noise. By noise, I mean randomised pixels at different brightnesses or colours to the desired signal. Noise tends to be an analogue artefact (e.g. grain in film or transmission noise ("snow") in an analogue tv picture). Digital compression systems are terrible at compressing noisy images because every pixel is different (look at this HBO logo as a good example).

16
  • Thank you so much for answering, I wasn't sure anybody would. According to this user who bought the exact same scanner and had one of his MP4 files analyzed (its the 5th post down) using an automated service, the scanner uses the AVC codec aka h.264. However, according to the link of the user manual you posted, if you look on page 3 of the PDF, under features, it simply says that it encodes using "MPEG-4." Sep 2 at 7:37
  • When the manual says "MPEG-4," do they really mean h.264 given that h.264 is sometimes called MPEG-4 Part 10? Sep 2 at 7:37
  • Video files have wrappers and codecs. The .mp4 part of the file name refers to the wrapper. You can have many different codecs inside a video file- eg different ones for audio and video tracks. But mp4 files most commonly use h264 as the codec.
    – tomh
    Sep 2 at 7:41
  • A lot of other things will affect the quality of the transfer, not least the optical quality of the signal path - eg the size and quality of the lens and the sensor. The bandwidth limitation of how the files are encoded will also affect the quality. Without having one to play with it’s tricky to offer any more advice.
    – tomh
    Sep 2 at 7:43
  • Help me understand your 4th paragraph. I thought MJPEG and MPEG-4 were two opposing compression standards, and so it wouldn't be advisable to use them both together, but the scanner's manual clearly states that the machine uses MPEG-4, with no mention of MJPEG, and in your reply you state that the scanner uses MJPEG, so which is it? Are you saying that this particular scanner uses both MJPEG and MPEG-4 compression together? Sep 2 at 8:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.