Jump to content

Attila Bakos

Members
  • Posts

    419
  • Joined

  • Last visited

Posts posted by Attila Bakos

  1. 8 minutes ago, Mokara said:

    The difference is due not to the interface itself, but what is happening internally in those devices. Just because an interface is capable of a particular speed does not mean that the device can generate or receive data at that rate. To use a UHS-II spec label they have to be capable of meeting the 300 mB/s data transfer requirement however. In the case of a camera like the NX, for example, there is a bunch of processing going on that limits the availability of the data, and that results in lower speeds when recording natively. However, we know the minimum base internal data rate of the camera since it does a 6.5k sensor read at 8 bits/30 fps in preparing data for 4K video. That is a bandwidth of approximately 630 mB/s. The camera does processing on that data however, and the bottleneck for data delivery is the processing itself, not the UHS-II interface. If you found a way to side step that processing you should be able to deliver enough data to swamp the interface, since the camera is dealing internally with at least twice the amount specified by UHS-II. If you only used a crop of the sensor however you would be able to (in theory) deliver a RAW data feed as I explained earlier.

    Yeah I get this and this is fine, but last time I checked the bitrate of 4K 12/14bit uncompressed RAW at 25fps exceeded the maximum continuous write speed of the fastest UHS-II cards. How do you solve that bottleneck?

  2. Just now, Trek of Joy said:

    Did you try it with something that has a little contrast and detail, like a human face, instead of a black leather chair with a set of black headphones draped over it? That looks bad, but your foreground subject with a blown white background gives the AF system little to lock on to as there's virtually no contrast.

    Chris

    Oh that's not my video, I don't have an X-H1 (yet?).

  3. Well, if you can truly record 4K uncompressed RAW at 25 fps in 12 or 14 bits, internally, that will indeed prove me wrong. If that happens, it will make a lot of people happy, myself included. Keep in mind though that I didn't make up things like bus speed or uncompressed raw bitrate requirements, so saying that I'm completely wrong without any facts is a bit meh. Prove me wrong for real :)

  4. 4 hours ago, Mokara said:

    If the camera has the UHS-II spec then it must be capable of that sort of transfer speed by definition. That is the whole point of a spec, if something complies with it, it has to meet it. If there is no processing going on then there should be nothing stopping the camera from achieving that.

    Yeah, "must be capable" is all we can say, but noone knows for sure (yet). Just check writing speeds of the best UHS-II card readers and even there you will see big differences. Somehow I seriously doubt that the NX1 will be as fast as those, but obviously I'm just guessing too, so let's say you're right, then we are looking at 8bit 4K RAW tops, for continuous recording without frame drops.

  5. On 2018. 03. 15. at 7:50 PM, Matthew Hartman said:

    That image sensor has a 14bit 6.5k RAW signal. 

    Exactly. That would require 1.14GB/s continuous write speed at 25fps. The NX1 has an UHS-II interface, the maximum bus speed is 312MB/s. But that's only theoretical, afaik the body can't even reach half of that. It's not even close to the requirement of UHD 14bit RAW at 25fps. 1080p maybe...

  6. 1 hour ago, Mattias Burling said:

    Yes, but it contains the same information. There are no crushed blacks thats not recoverable with a simple preset. After a quick adjustment (one click) I see pretty much the same image. What Im trying to determine is if the -2 shadow gives you information that otherwise is lost. Because that was what I doubted from the beginning and that spurred this hole thing :)
     

    I have already proven that they do not contain the same information when I posted my frame grabs (on page 51) where I pushed the shadows to the same level, and you could clearly see that the -2 version has more detail and less color noise in the darkest parts. Now that you have the sources you can easily replicate that.

    But here is another proof for you, I cut down the bottom 694 pixels of the images, so that we only see the darkest parts:

    This is the 0/0 version, it has 639 unique colors, and the green channel has 40 unique values.

    cut_0.thumb.jpg.29902b2a265d488ff615bd72e0027636.jpg

    This is the -2/-2 version, it has 1137 unique colors, and the green channel has 57 unique values.

    cut_-2.thumb.jpg.ffab204c9897abc0c0e43b407db569cf.jpg

    The framing is not perfectly the same but it wouldn't make a difference. You are working with less shadow information in the 0/0 version.

  7. 2 minutes ago, Mattias Burling said:

    Cool, thanks :) 
    When quickly comparing shadows -2 and shadows 0 I see no difference but I will have to take a better look later when I get home.
    Just to make sure, these where exact same aperture/shutter/iso/ND/etc? Identical exposure?

    I shot these almost a year ago, so I can't be 100% sure, but I believe exposure is identical. I'm not sure why you don't see a difference, to me it's clearly there:

    0/0:

    highlights_0_shadows_0.thumb.jpg.7bc48cd6d71a1d37ce98c2ffff83a51b.jpg

    -2/-2:

    highlights_-2_shadows_-2.thumb.jpg.af36ce6cf5cb3a59301bb701e3ff4275.jpg

  8. 22 minutes ago, Mattias Burling said:

    Could someone with the camera shoot a second or two with for example Eterna in default and then the same scene with lifted shadows and different DR setting. And then share the clips?
    Would be interesting to see if there truly is a difference or if its replicable in post.

    Remember the comparison I showed you earlier? I found the sources for it: link It's not exactly what you want, but it shows that the highlights and shadows settings make a difference you can't replicate in post if your scene is very contrasty. Remember to switch the clips to full range, if your NLE doesn't detect it as such.

  9. @Mattias Burling Regarding your comparison, there's also a possibility that your NLE interpreted the X-T20 footage as video range. In this case switch it to full range manually, and see if that makes a difference. Check the Shadow & Highlight tones part in this review: link That's the difference you want to see in video mode. But this is pretty much offtopic here, we can continue in PM if you want to.

  10. 1 minute ago, Mattias Burling said:

    Cool, can you share the original clips or perhaps quickly shoot two new once?
    Would love to try again and see if I can see a difference in post.

    Sorry I don't have the originals anymore, and I also sold my X-T2 a few weeks ago. But this should be easy to replicate, if you have a Fuji now.

  11. 2 minutes ago, Mattias Burling said:

    I believe you :) 
    Im just talking about my needs and experiences. Ive never challenged anyone, thats just something the angry fellow makes up in his head. Must be the sun down there.

    I see no difference between them worth mentioning. I see zero reason to play with shadows and highlights in camera since I have been able to replicate it in post.
    That doesn't mean you didn't see a difference. We simply might have been under different circumstances, expose differently, or edit differently.
    Or we just have different bars when it comes to what we view as a significant difference.

    Its just like when I said Im glad they stuck with a flip screen, because I have much more use for that than a fully articulating. Some would just chalk it up under "a different preference". And a few angry fellows will go ape shit and try everything to ridicule it, convince me its "wrong" or just act up like a baby.

     

    That's totally fine by me :) But since we are talking about it, I searched through my folders and found one example. Here I didn't touch highlights, only shadows, shot the scene with shadows set to 0, then shadows set to -2, and pushed both to the same level in post. While the difference is not that much, it's there. Check the grass (in full screen) in the lower right section, more details and less color noise at -2.

    Shadows at 0:

    pronegstd_0_shadows.thumb.jpg.9987bdf95a6e72b23791f33b6e2057ce.jpg

    Shadows at -2:

    pronegstd_-2_shadows.thumb.jpg.e9bd1b3dcfaac52399ff10ea185b6e19.jpg

  12. 14 minutes ago, Mattias Burling said:

    The shadows/highlights settings doesn't really do anything different.

    But they do, especially shadows. I did many tests in the past, the difference is visible in high contrast situations. There was a discussion about this on liftgammagain where I posted my results but unfortunately I deleted them long time ago and I don't have a Fuji body at this moment. So you either believe me or not :)

  13. I love Fuji's philosophy about updates and I loved my X-T2 very much, however, webrunner5 is right that they rushed the X-H1. They aimed this body partly at videographers, and you don't want do this with such a flaw in CAF. You also don't want to do this with sudden and therefore very visible corrections in your video while using IBIS, which is a major selling point for video. It seems to be calibrated for photo mode and I think it should work differently in video mode. I'm no expert there, but I'm certainly not the first who noticed this. 120p has oceans of moiré/aliasing, this is something I doubt they can fix, but what about 1080p60? There was a comparison earlier in this thread that shows lesser quality 1080p60 compared to the X-T2. It would be great if someone could confirm that too.

×
×
  • Create New...