Thread: Best article on resolution ever

Page 1 of 2 12 LastLast
Results 1 to 10 of 17
  1. #1 Best article on resolution ever 
    Senior Member Frank Glencairn's Avatar
    Join Date
    Apr 2012
    Location
    Germany
    Posts
    3,764
    title says it all - have fun guys

    https://ascmag.com/articles/a-clear-...-of-resolution
    Reply With Quote  
     

  2. #2  
    Senior Member
    Join Date
    May 2012
    Location
    Germany
    Posts
    924
    The guy is super cool. His film vs digital comparison is just as telling.

    https://www.youtube.com/watch?v=suZtYPIADHM
    Reply With Quote  
     

  3. #3  
    Senior Member Abobakr's Avatar
    Join Date
    May 2012
    Posts
    662
    wow.. such informative
    Reply With Quote  
     

  4. #4  
    Member
    Join Date
    Jul 2017
    Location
    austin,tx
    Posts
    34
    very good info.. thanks for the share! showed me alot..
    Reply With Quote  
     

  5. #5  
    Senior Member Timothy Cook's Avatar
    Join Date
    May 2014
    Location
    Ft. Worth, Texas
    Posts
    841
    I hadn't seen the blog write up, thanks Frank. Cool at the end of the blog he writes, that all the test won't matter soon enough because these cameras too will be obsolete.

    While we/he are/is on the subject of resampling footage... here are my examples and reasons for resampling my Pocket footage from 1920 to 3840 in Resolve. If trying to make the determination with these shots on a phone I don't think you'll see it.
    All of these images where shot on my Pocket Cinema (R.I.P.) https://drive.google.com/drive/folde...0U?usp=sharing. NONE OF THESE SHOTS HAVE BEEN SHARPENED OR BLURRED, JUST COLOR GRADED ONLY.. Feel free to download the tiff files and compare at home.

    "Girl On Bed" with both tiff files open full screen on a 5K monitor they look identical, minus the upsampled shots just look cleaner. Will any of this be noticed on a giant screen? Or on a high end high resolution monitor? Some have eyes that work better than others so who knows.

    On the 1920 sampled shoots (originals), once I zoom in by a factor of one, I start to slightly notice the pixels around the edge of objects.
    Her Arms, the dog's nose, the dark lines under the dog's tear ducts all have large size pixel trying to make edges.

    Zoom in by a factor of two or three and you start to see the aliasing in the hair pixels on every edge, aliasing between the freckles on her left arm, etc etc.

    All of these are absent on the upscaled image (original upscaled to 3840 in Resolve), until you zoom in to a factor of six or seven times. And even then the corners and edges look a lot cleaner. Noise cleans up too.

    Same scenario with the Door shots.

    I can only imagine if I would have shot these scenes on the UMP 4.6. Or maybe Resolve's algorithm for upscaling footage is really really good.

    I love that this DP is taking the time to explain and educate with his years of experience. He definitely may be onto something that 4K maybe the magic number, cause even my Pocket footage upsampled to 4k starts to look pretty dang good. But I've never seen it on the big screen before.
    Last edited by Timothy Cook; 07-25-2017 at 07:18 PM.
    Reply With Quote  
     

  6. #6  
    Senior Member
    Join Date
    Sep 2012
    Posts
    1,735
    So the two takeaways I got beyond more pixels aren't necessarily better;

    1. The Alexa 65 looks better than an 11k Imax scan or by extension any camera on the planet.

    2. The scaling algorithm is pretty important, which begs the question.

    Does Resolve always use the same scaling algorithm, and are there better options?

    There are four places to scale footage in Resolve. Import e.g. 2K into a 4K timeline, edit tab transform, color tab sizing, and deliver tab. Does anyone have any data or information that supports these all use the same scaling algorithm or is it different? Is it better to use one over the other? I've personally tested 4K camera footage shot to 2K that looked better on a 4K timeline than it did on a 2K timeline, conversely I've shot 4K footage zoomed in on a 2K timeline and it got pretty soft by about 130% using the edit transform. Basically what is the best practice for scaling footage within Resolve? Is it better to scale outside of Resolve?
    Reply With Quote  
     

  7. #7  
    Senior Member Timothy Cook's Avatar
    Join Date
    May 2014
    Location
    Ft. Worth, Texas
    Posts
    841
    Quote Originally Posted by Howie Roll View Post
    So the two takeaways I got beyond more pixels aren't necessarily better;

    1. The Alexa 65 looks better than an 11k Imax scan or by extension any camera on the planet.

    2. The scaling algorithm is pretty important, which begs the question.

    Does Resolve always use the same scaling algorithm, and are there better options?

    There are four places to scale footage in Resolve. Import e.g. 2K into a 4K timeline, edit tab transform, color tab sizing, and deliver tab. Does anyone have any data or information that supports these all use the same scaling algorithm or is it different? Is it better to use one over the other? I've personally tested 4K camera footage shot to 2K that looked better on a 4K timeline than it did on a 2K timeline, conversely I've shot 4K footage zoomed in on a 2K timeline and it got pretty soft by about 130% using the edit transform. Basically what is the best practice for scaling footage within Resolve? Is it better to scale outside of Resolve?

    Howie, with my two examples above, I started the project off, in Project Settings, at 3840 and ran it that way all the way through. (I finally have a work station that can handle that). But I have also just rescaled other projects to 3840 resolution at delivery and both 4K files looked identical, if I had upscaled at the beginning or the end. I have noticed that I can change the project setting in mid project back and forth without any anomalies happening.

    Someone more knowledgable could probably confirm or debunk my workflow though.
    Reply With Quote  
     

  8. #8  
    Senior Member Asyndeton's Avatar
    Join Date
    Feb 2016
    Location
    Portland, OR
    Posts
    199
    Quote Originally Posted by Timothy Cook View Post
    Howie, with my two examples above, I started the project off, in Project Settings, at 3840 and ran it that way all the way through. (I finally have a work station that can handle that). But I have also just rescaled other projects to 3840 resolution at delivery and both 4K files looked identical, if I had upscaled at the beginning or the end. I have noticed that I can change the project setting in mid project back and forth without any anomalies happening.

    Someone more knowledgable could probably confirm or debunk my workflow though.

    This thread is a good source of information about Resolve's input and output sizing: http://www.liftgammagain.com/forum/i...17/#post-46663

    Also want to note that if your source footage is 4K 16:9 and you set your timeline resolution to 1080 16:9, then render out your footage at 4K 16:9, you're essentially 'up-scaling' your footage. So you want to make sure your timeline resolution matches whatever your final deliverable should be. You usually won't notice the difference unless you A/B the images back-to-back, but it will be slightly softer if you're timeline resolution is down-scaling the footage and you up-scale it again while rendering out. My normal workflow is to make my timeline resolution the 1080 equivalent of the source footage's aspect ratio to be less taxing on the machine and change it to the final deliverable's specs when delivering. Any windows and tracking information will properly scale as long as it's the same aspect ratio.

    That said, I've been enjoying Yedlin's tests as they're super educational. His older Alexa vs 35mm test, the footage between the two is nearly indistinguishable.
    Last edited by Asyndeton; 07-26-2017 at 01:03 AM.
    Reply With Quote  
     

  9. #9  
    Senior Member
    Join Date
    Sep 2012
    Posts
    1,735
    Hmmm... It's interesting, when I have 4k footage on a 2K timeline even in Multicam mode with some mild color tab work the footage plays smoothly on my new MacBook. When I do the reverse and have 2K footage on a 4K timeline the footage plays back choppy even if it's just a single test clip. It seems as if the upscaling algorithm is more complex/computatioanally difficult where if it were the same algorithm it would just be / vs X. My gut feeling is that resolve treats higher res footage more sloppily because it can and is a little more tender when uprezzing because it needs to be. I tested some footage here http://www.bmcuser.com/showthread.ph...S-UHD-timeline and concluded that I'd get best results shooting 2K in a 4K timeline as counterintuitive as it seems but that admittedly is just for my camera.
    Reply With Quote  
     

  10. #10  
    Senior Member
    Join Date
    May 2012
    Location
    Germany
    Posts
    924
    I'd wish for very good upscale algorithms in Resolve, something like the After Effects' detail-preserving upscale. Even as it is right now though, rendering HD/2K sources to 4k masters gives much better results when played back in HD. Not sure whether the moment of upscaling during workflow matters.
    Reply With Quote  
     

Similar Threads

  1. Great article about 4.6k UM
    By renegadewill in forum General Discussion
    Replies: 0
    Last Post: 11-10-2016, 01:37 PM
  2. Fangoria did an article on my film.
    By cineman in forum General Discussion
    Replies: 6
    Last Post: 08-20-2014, 10:38 AM
  3. American Cinematographer article
    By scorsesefan in forum Off Topic
    Replies: 1
    Last Post: 05-23-2014, 03:31 PM
  4. Article: Why You Can't Use S16 Lenses with BMCC
    By Charlie Doom in forum Lens Discussion
    Replies: 3
    Last Post: 01-10-2013, 12:08 AM
  5. ARTICLE: BMC - First Impressions
    By Stephen Mick in forum General Discussion
    Replies: 14
    Last Post: 04-27-2012, 09:24 PM
Bookmarks
Bookmarks
Posting Permissions
  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •