Stories
Slash Boxes
Comments

SoylentNews is people

posted by takyon on Friday December 28 2018, @10:09AM   Printer-friendly
from the mirror-mirror dept.

Submitted via IRC for Fnord666

2018 was the biggest shakeup in years for the camera world

2018 was the tipping point for mirrorless cameras.

If you're a photographer who fears change, 2018 might've shook you up. First Sony launched the A7 III, arguably the world's best full-frame camera, then Fujifilm released the X-T3, the top APS-C model you can buy right now. Right after that, Canon and Nikon launched all-new full-frame mirrorless systems with three new cameras, the EOS R, Z6 and Z7. To top it off, mirrorless video champ Panasonic announced it was diving into full-frame mirrorless as well with two new models, the S1 and S1R.

This is the biggest upheaval in the camera industry for years and could have a big impact on your buying decisions. On top of that, companies that don't adapt quickly may not survive, especially in a market gutted by ever more incredible smartphone cameras -- and moving fast will be a challenge for conservative companies like Canon and Nikon. Based on everything that happened in 2018, you can expect more drama and turmoil in 2019, but also even more innovative and interesting cameras.

[...] While Full-frame is great, most of us don't have $1,500 or more to throw down on a camera, let alone the lenses. What I hope to see is the same level of innovation on more affordable mirrorless products that cost under $1,000. With multi-camera AI-powered smartphones starting to close the gap there, too, camera companies have got to bring some of the same capabilities. Here's hoping that 2019 is just as eventful as 2018, if not more so.


Original Submission

 
This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 2) by fyngyrz on Saturday December 29 2018, @03:42PM (3 children)

    by fyngyrz (6567) on Saturday December 29 2018, @03:42PM (#779705) Journal

    Here's the problem with your reasoning. You talk about a tiny sensor not capturing enough light. How about at ISO200? Oh, it'll get blurry? Well how about tiny sensor, iso 10k, 100 shots. This lets me know what moved, and apply light collection for the same object across different sensors/pixels. Bam, ISO200 that's sharp and noise-free. This is what they meant by AI.

    There's no problem with my reasoning. But there's a problem with yours; it come down to physics.

    Whatever a tiny sensor can do in 100 shots, a large sensor can do better in fewer shots.

    The larger sensor, together with its larger lens, gathers (a lot) more light, faster, at more advantageous signal to noise levels.

    You're thinking that stacking (which is all the AI process you're talking about does, it just does it smarter) provides a gain (and it does) but you didn't think about the general consequences of that. Here's the bottom line:

    Any technique that can be applied to the data gathered by a tiny sensor can be applied to the data gathered by a large one.

    The large sensor has inevitably done the data gathering better. When designed, the large sensor also has two ways it can be optimized to use that additional surface area. It can be set up with larger wells, which improves the signal to noise ratio, or it can be set up to pack in more wells overall, which increases the spatial resolution at the expense of noise. For the former, you get faster, higher quality data. For the latter, you get more detail, which can be driven even to the point of pushing current lens technology right to the edge. Typically, a compromise is made so that you get some of both: higher spatial resolution and improvement in signal to noise ratio. The result, no matter what the choices are here, is better image data per image acquisition.

    The issue is that doing whatever with data inevitably works better, when the data is better.

    For instance, 100 shots at 1/60th of a second takes six seconds. If you're trying to shoot something that's moving, that may very likely require that you physically track it with the camera or it will go out of frame. While you track it, the amount of data gathered at the edges of the image will vary because your tracking can't be perfect; that will have negative consequences for the ultimate result.

    Compare this to a sensor that can acquire the same shot in a single sample: it takes 1/60th of a second, and all the resulting data is of pretty much similar quality, so the edges aren't compromised — there are no poorly tracked regions, because there was no tracking going on.

    Another thing: with region-variable stacking, as a tracked object moves, the newly exposed detail revealed by the motion of the objects moving within the frame is gathered fewer times (because it was previously covered.) This again means we have regions with lower-quality data, only this time, they are not at the edges, but instead are adjacent to the objects moving within the frame.

    Yes, algorithms can definitely help. You don't have to convince me of that; I write those algorithms for a living. But no, there isn't any magic that will help smaller sensors more than larger sensors.

    --
    Physics: There are some laws you can't break.

    Starting Score:    1  point
    Karma-Bonus Modifier   +1  

    Total Score:   2  
  • (Score: -1) by fakefuck39 on Monday December 31 2018, @07:40AM (2 children)

    by fakefuck39 (6620) on Monday December 31 2018, @07:40AM (#780178)

    So, I stopped reading here: "Whatever a tiny sensor can do in 100 shots, a large sensor can do better in fewer shots."

    You're a retard. You took the claim of the article, which said phone cameras are starting to overlap with the domain of nice cameras, and called that invalid. I stated my case - a case for 90% of the people who had good cameras, and explained why I, and most others, now use a phone camera. I don't care about a larger sensor doing it with fewer shots or one shot. The domain my use case for a pro camera was in is now covered by the phone. Article claim true. The rest of your logic is simply changing the subject, and I'm not here to chat with you. I'm just here to tell you you're a retard because your warped brain chooses to ignore very basic facts in front of you.

    • (Score: 2) by fyngyrz on Monday December 31 2018, @04:45PM (1 child)

      by fyngyrz (6567) on Monday December 31 2018, @04:45PM (#780301) Journal

      So, I stopped reading here

      And I stopped reading right there. Cheers.

      --
      Reality is that thing which, when you stop believing in it, doesn't go away.

      • (Score: -1) by fakefuck39 on Tuesday January 01 2019, @03:18AM

        by fakefuck39 (6620) on Tuesday January 01 2019, @03:18AM (#780512)

        you're not my audience. you're my personal clown - with that combination of retard and purposefully pretending to be dense. i don't give a fuck if you read what I wrote - I'm surprised you can read at all.