Spencer Hutchins Posted November 19, 2008 Share Posted November 19, 2008 So being from a film school, I often hear of these important instances that students (people who know nothing) freely throw around their 'genius' to each other on topics. So with that being said, I am usually confused: When your final project will be going SD, does shooting HD and down-converting to SD indeed give you some kind of better quality, or is SD in fact SD no matter where it comes from? Again, sorry for having to ask these questions but thank you for your time. Link to comment Share on other sites More sharing options...
Premium Member Chris Keth Posted November 19, 2008 Premium Member Share Posted November 19, 2008 No advantage that I know of, unless it's for effects. Doing effects on a greater format then downconverting them often makes them look cleaner. There's also the possibility of the HD format having better colorspace than the SD format in question, but that's not a set rule or anything. It depends on which HD and which SD formats in question. Perhaps they just want to be able to display the work in HD as well as SD? Link to comment Share on other sites More sharing options...
Premium Member John Sprung Posted November 20, 2008 Premium Member Share Posted November 20, 2008 HD downconverted to SD gives you more resolution than is possible in native SD. The reasons involve the Nyquist limit and the optical low pass filters of the cameras. To prevent aliasing, Nyquist tells us that we have to filter out all detail finer than twice the distance between photosites. So, if we have N photosites across the chip, the largest number of lines we can reproduce would be N/2 across the whole picture. To do that optically, the filter has to start rolling off at N/4. Downconversion from HD to SD is a re-sampling which also requires a filter to make the SD Nyquist limit. But this being done digitally, it doesn't have the severe restriction of an analog optical filter. Digital filters can use negative numbers, but the optical filter doesn't get to use anti-photons. So, the downconversion can make the SD Nyquist limit using a "brick wall" filter. That gives it a lot more top octave sharpness than native SD. -- J.S. Link to comment Share on other sites More sharing options...
Premium Member Chris Keth Posted November 20, 2008 Premium Member Share Posted November 20, 2008 Touché, John! It makes total sense now that's it's explained to me so well. I learn something here every day, thanks. Link to comment Share on other sites More sharing options...
Spencer Hutchins Posted November 21, 2008 Author Share Posted November 21, 2008 Great Information, both John and Chris!! ;) Thanks guys Link to comment Share on other sites More sharing options...
Premium Member Michael Nash Posted November 21, 2008 Premium Member Share Posted November 21, 2008 John's absolutely correct about the oversampling issue, but pixel count isn't all of it. The other factor is the quality of the SD and HD cameras you're comparing. A broadcast-level SD camera with a good lens can deliver an exquisite SD image, while the downconverted image from a consumer-grade HD camera may look only so-so. It's not just the size of the image; it's the quality. Link to comment Share on other sites More sharing options...
Spencer Hutchins Posted November 21, 2008 Author Share Posted November 21, 2008 Good call Michael, can't forget lenses always provide more clarity to the images. Link to comment Share on other sites More sharing options...
DJ Joofa Posted November 21, 2008 Share Posted November 21, 2008 John's absolutely correct about the oversampling issue, but pixel count isn't all of it. The other factor is the quality of the SD and HD cameras you're comparing. A broadcast-level SD camera with a good lens can deliver an exquisite SD image, while the downconverted image from a consumer-grade HD camera may look only so-so. It's not just the size of the image; it's the quality. Another important consideration is the quality of downsampling process itself. With the advent of 4K and higher resolution and the need to downsample it to SD/HD sized footage, care must be taken in the resampling process. It appears many software and hardware solutions out there, including publications in journal and conferences have gotten some understanding incorrect, and even when understanding was correct, some mistakes in the implementation. (Many open source software seem to have some aspects of implementation wrong. The results still look very good on certain downsampling ratios, but don't work on others, with the result that authors claim erroneous conclusion regarding what filters work best and when. An article describing a fast hardware solution for fast downsampling in a very prestigious IEEE publication got the implementation wrong so that it actually boiled down to a simple moving average process.) Link to comment Share on other sites More sharing options...
James Martin Posted January 27, 2009 Share Posted January 27, 2009 For me, the proof is in the pudding. If what you hypothesise were true, then Digibeta would be the equal of 65mm (on DVD anyway). Of course, it isn't. One must also remember that we're at a transition period right now and though SD is acceptable for broadcast today, your program could be excluded from a tasty sale in the future if you don't have an HD master to fall back on! Link to comment Share on other sites More sharing options...
Premium Member Walter Graff Posted January 28, 2009 Premium Member Share Posted January 28, 2009 It all sounds good... I shoot everything on various forms of HD from f950s to JVC HDV that I end up editing and outputting to SD medium. No advantage for the most part other than being able to zoom in more during edit in an SD timeline if I end up doing it that way. http://www.bluesky-web.com/sdhd.html Link to comment Share on other sites More sharing options...
Premium Member Alessandro Machi Posted February 3, 2009 Premium Member Share Posted February 3, 2009 HD downconverted to SD gives you more resolution than is possible in native SD. The reasons involve the Nyquist limit and the optical low pass filters of the cameras. To prevent aliasing, Nyquist tells us that we have to filter out all detail finer than twice the distance between photosites. So, if we have N photosites across the chip, the largest number of lines we can reproduce would be N/2 across the whole picture. To do that optically, the filter has to start rolling off at N/4. Downconversion from HD to SD is a re-sampling which also requires a filter to make the SD Nyquist limit. But this being done digitally, it doesn't have the severe restriction of an analog optical filter. Digital filters can use negative numbers, but the optical filter doesn't get to use anti-photons. So, the downconversion can make the SD Nyquist limit using a "brick wall" filter. That gives it a lot more top octave sharpness than native SD. -- J.S. Does this apply to every kind of shot, or does it vary. For instance, a wide, backlit shot may look superior if downconverted, but what if the shot is a close-up? Link to comment Share on other sites More sharing options...
Premium Member Keith Walters Posted April 18, 2009 Premium Member Share Posted April 18, 2009 It all sounds good... I shoot everything on various forms of HD from f950s to JVC HDV that I end up editing and outputting to SD medium. No advantage for the most part other than being able to zoom in more during edit in an SD timeline if I end up doing it that way. http://www.bluesky-web.com/sdhd.html All I can say is that you must use some very ordinary studio equipment. Even back in the days of analog HD it was easily demonstrated that downconverted HD would give a superior picture to that obtained from an otherwise equivalent SD camera. The reality is, a 720 x 576 pixel (or 640 x 480) SD camera can only practically reproduce half that number of actual lines on a test chart, due to interlace and aliasing effects. If the picture is 2X oversampled, there are major benefits: (As John pointed out) The OLPF can have a much less savage roll off. And then, because the SD LPF function can be carried out post-capture in software, the downconverting software gets a chance to look at how the image comes out both, with and without, pre-filtering. The thing is, most of the time, large parts of the image don't actually need the OLPF; it's really only necessary when there are areas of repetitive fine detail which produce Moire patterning. Even though sharp single edges theoretically violate the Nyquist limit, in practice the effect this produces is far less objectionable than moire patterning, and the added sharpness they add to the image more than compensates for this. However these sharp edged "babies" tend to get thrown out with the moire "bathwater" when a hardware OLPF is used. With a software filter, the filtering only needs to be applied where there are visible artifacts, and the degree of filtering can also be controlled. A similar thing applies with interlace. Because the image is not set in SD "stone" it's possible for the downconverting software to detect and minimize the few places where interlace artifacts are most noticeable without killing the overall vertical resolution. The overall result is that with the right source material and a good analog decoder, standard definition TV on a good interlace-free HD display can produce image quality approaching all that was ever expected from interlaced-scan analog HD on CRTs in the early 1990s. Which is one argument for 4K acquisition for 1920 x 1080 HD. Link to comment Share on other sites More sharing options...
Recommended Posts
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 accountSign in
Already have an account? Sign in here.
Sign In Now