3 ms·

I've been using a professional camera (Sony A7C) for a bit over two years now, and my wishlist looks very different than yours. - Modern computational photogra

I've been using a professional camera (Sony A7C) for a bit over two years now, and my wishlist looks very different than yours. - Modern computational photography: Imagine having access to a high-end sensor combined modern computational photography tooling. Smartphones do incredible things with computational photography, including HDR, low light stacking, noise reduction, upscaling, and picking out smiling faces from multiple photos. Most of these things can be done in post on a professional camera with tools like Photoshop, but often they are inferior, using a single image to worse affect, requiring a lot of manual work, or just having less mature tooling. I'd love to have access to the raw data and computational pipelines to selectively apply different processing and tweak computational settings in post. - Better UI: Sony notoriously has a terrible UI in their camera menu systems. Working with it feels like something I would've used on a flip-phone from 20 years ago, with things buried multiple levels down in obscure menus and vague, sometimes poorly translated descriptions. The hardware is great, but the GUI feels like something built by engineers rather than for end-users.

atoav 2y ago
Sony is the manufacturer who had me search for the battery release camrra for a solid minute - it was on the opposite side of the camera. If I didn't have to use those cameras they would be engineering marbels, but their professional video segment really could use a new generation of designers.
dannyw 2y ago
Most computational photography inherently breaks giving the photographer direct manual control over how a picture is taken (shutter speed, aperture, ISO). For example, HDR or low light staking means taking multiple photos. This requires a fast shutter speed to minimise change in the scene or subject. However, photographers sometimes intentionally want a slower shutter speed for motion blur, or to shoot at lower ISO for less noise and more resolution. Actions like noise reduction, upscaling, etc are post-processing actions. They are best done on a large, high-resolution, color-calibrated screen; instead of a portable device with a small battery where you want to maximise the number of shots you can take before swapping or recharging batteries. Bear in mind that modern cameras easily shoot in 40, 60+ MP. To truly perform noise reduction, or upscaling on those images require a lot more processing power, even with fixed-function hardware. Your "48MP" quad-bayer smartphone is generally processing a "pixel-binned" 12MP image straight out of the sensor, that might be resampled to 24MP (iOS) after post-processing; but was never anything more than 12MP. The hardware was never more than 12MP, except they sub-divided each bayer pixel into 4 quad (or 12 for some Samsung chips) to claim 4x/12x the spec sheet w/o real world benefits. Your smartphone's SoC, is usually manufactured on the latest, bleeding edge 3nm or 4nm processes. They are also the most expensive; made practical because tens of millions of each chip are shipped each year. There are ~1m dedicated cameras shipped each year; from many different manufactures using different chips; usually built on N16, N12, or N7 for the bleeding edge ones. No, they cannot just use a smartphone SoC. Smartphone image signal processors simply can't process 61MP at 20FPS or 30FPS or whatever the burst rate sports photographers demand. Your smartphone uses an electronic shutter. Your proper camera probably has a mechanical shutter, which is generally superior but also requires more force to move. The battery in your smartphone is probably ~2x bigger than a professional mirrorless or DSLR. Pro photographers buy cameras, in a significant part based on how many shots they can take. > the GUI feels like something built by engineers rather than for end-users. Think of the GUI as being built for professional photographers who's been using an UI for the past 20 years of their livelihood. You would probably be really upset if your OS decided to change established keyboard shortcuts like Ctrl+C or Ctrl+V. You probably wouldn't choose that OS, and if professional photographers won't buy a camera because the UI is "weird to them", you've wasted your R&D costs because the dedicated camera market for the past decade is essentially only professionals; with some amateurs and hobbyists.
- Arcanum-XIII 2y ago
  Everything you say is pretty much on target, except the UI. I know my Sony camera, and every time I need to do a change in a setting, it's plainly horrible. Confusing naming, no feedback, sometimes you're thrown back to the main UI without explanation... and the bugs. For the longuest time the auto off was not working properly, which, on an hybrid camera leads to a very short battery life. From what I've seen, it's not better on Canon and Nikon hardware. They spent a lot of time refining the manual control, so it's very good and intuitive but then fails so bad at the electronic UI...
starky 2y ago
Exactly, the modern mirrorless camera should be providing options to use the latest technology as well as the full manual control for when you need it. Imagine what you could do with the stacking capabilities of modern smartphones with a stacked image sensor like on the Sony A9 series that can essentially take full resolution photos at video speed. Relying solely on image processing isn't be a feasible product. Artists want full control to create the image that they visualized, not just what some ML algorithm determined was best, but at the same time there are many situations where you can gain improved image quality that these features can accomplish. Additionally, cameras have to support unedited images as a large customer base is journalism that requires minimal to no editing be done on images. The UI thing is massive. I'm surprised nobody has created a UI more similar to the Hasselblad one that looks good and shows your settings very clearly. It would be great if the interface you interacted with was clear and looked like a modern UI, but still had the plethora of options available under the hood in an organized and quick to access manner. You don't often need to change many options, but when you do they need to be easy to find.

atoav 2y ago

Sony is the manufacturer who had me search for the battery release camrra for a solid minute - it was on the opposite side of the camera. If I didn't have to use those cameras they would be engineering marbels, but their professional video segment really could use a new generation of designers.

dannyw 2y ago

Most computational photography inherently breaks giving the photographer direct manual control over how a picture is taken (shutter speed, aperture, ISO). For example, HDR or low light staking means taking multiple photos. This requires a fast shutter speed to minimise change in the scene or subject. However, photographers sometimes intentionally want a slower shutter speed for motion blur, or to shoot at lower ISO for less noise and more resolution. Actions like noise reduction, upscaling, etc are post-processing actions. They are best done on a large, high-resolution, color-calibrated screen; instead of a portable device with a small battery where you want to maximise the number of shots you can take before swapping or recharging batteries. Bear in mind that modern cameras easily shoot in 40, 60+ MP. To truly perform noise reduction, or upscaling on those images require a lot more processing power, even with fixed-function hardware. Your "48MP" quad-bayer smartphone is generally processing a "pixel-binned" 12MP image straight out of the sensor, that might be resampled to 24MP (iOS) after post-processing; but was never anything more than 12MP. The hardware was never more than 12MP, except they sub-divided each bayer pixel into 4 quad (or 12 for some Samsung chips) to claim 4x/12x the spec sheet w/o real world benefits. Your smartphone's SoC, is usually manufactured on the latest, bleeding edge 3nm or 4nm processes. They are also the most expensive; made practical because tens of millions of each chip are shipped each year. There are ~1m dedicated cameras shipped each year; from many different manufactures using different chips; usually built on N16, N12, or N7 for the bleeding edge ones. No, they cannot just use a smartphone SoC. Smartphone image signal processors simply can't process 61MP at 20FPS or 30FPS or whatever the burst rate sports photographers demand. Your smartphone uses an electronic shutter. Your proper camera probably has a mechanical shutter, which is generally superior but also requires more force to move. The battery in your smartphone is probably ~2x bigger than a professional mirrorless or DSLR. Pro photographers buy cameras, in a significant part based on how many shots they can take. > the GUI feels like something built by engineers rather than for end-users. Think of the GUI as being built for professional photographers who's been using an UI for the past 20 years of their livelihood. You would probably be really upset if your OS decided to change established keyboard shortcuts like Ctrl+C or Ctrl+V. You probably wouldn't choose that OS, and if professional photographers won't buy a camera because the UI is "weird to them", you've wasted your R&D costs because the dedicated camera market for the past decade is essentially only professionals; with some amateurs and hobbyists.

Arcanum-XIII 2y ago
Everything you say is pretty much on target, except the UI. I know my Sony camera, and every time I need to do a change in a setting, it's plainly horrible. Confusing naming, no feedback, sometimes you're thrown back to the main UI without explanation... and the bugs. For the longuest time the auto off was not working properly, which, on an hybrid camera leads to a very short battery life. From what I've seen, it's not better on Canon and Nikon hardware. They spent a lot of time refining the manual control, so it's very good and intuitive but then fails so bad at the electronic UI...

starky 2y ago

Exactly, the modern mirrorless camera should be providing options to use the latest technology as well as the full manual control for when you need it. Imagine what you could do with the stacking capabilities of modern smartphones with a stacked image sensor like on the Sony A9 series that can essentially take full resolution photos at video speed. Relying solely on image processing isn't be a feasible product. Artists want full control to create the image that they visualized, not just what some ML algorithm determined was best, but at the same time there are many situations where you can gain improved image quality that these features can accomplish. Additionally, cameras have to support unedited images as a large customer base is journalism that requires minimal to no editing be done on images. The UI thing is massive. I'm surprised nobody has created a UI more similar to the Hasselblad one that looks good and shows your settings very clearly. It would be great if the interface you interacted with was clear and looked like a modern UI, but still had the plethora of options available under the hood in an organized and quick to access manner. You don't often need to change many options, but when you do they need to be easy to find.