Budget Camera Recomendation

[paulca]

My theory is that you can always make it smaller, but you cannot make it bigger. For example, I record audio at 24/48 into separate channels. After mastering, I render it to 16/44.1 stereo for playback. I am going to have to process an image anyway; crop, resize, whatever, so why not start with more than I need.

This kinda happened in video world.  While everyone was using 1080p pro video cameras where 4K.  This allows them to pan and scan the picture to pull out a native resolution sub frame when editing/mastering.  It also allowed for special effects and image stablising room.  Then everyone wanted 4K TV, so camera had to go up to recording in 5-8K.  Now there is talk of pushing 8K for home use in the next 5 years.  I don't know if there even are commercially available 12-16K video cameras!

[David Hess]

My theory is that you can always make it smaller, but you cannot make it bigger. For example, I record audio at 24/48 into separate channels. After mastering, I render it to 16/44.1 stereo for playback. I am going to have to process an image anyway; crop, resize, whatever, so why not start with more than I need.

Where this breaks down is that increasing the sensor resolution decreases the pixel size and aperture compromising performance so the result is not quite the same, but of course it is almost always acceptable.  More relevant may be that the existing optics are almost always limiting performance of the lower resolution sensor so using a fraction of the higher resolution sensor image increases the various optical problems.  On close examination, this is readily apparent on my point-and-shoot Canon Powershot where visible coma and chromatic aberration become evident with high contrast sources.

Marketing can make good use of an objective but largely meaningless specification like number of pixels while quantifying something which is more important, like the various forms of image distortion produced by the optics, is difficult.  Sensor resolution passed the bounds of diminishing returns long ago.

[magic]

Ironically, phones could have an edge over point and shoots because they use prime lenses which are easier to optimize and make with wide aperture. I mean, it is an edge as long as you are OK with whatever angle of view the manufacturer chose for you But anyway, it seems that about f/2 is the norm for phones these days, which enables a few megapixels worth of detail on a 1/3"~1/2.5" sensor if aberrations are taken care of.

That being said, I stumbled upon a review of some super-duper iPhone 11 Pro Max today, apparently Apple's TOTL phone, and I am absolutely not impressed by their sensor / image processing performance. A bunch of sample photos shot in bright sunlight and every single one has painful noise reduction artifacts all over the place. Somehow small sensor p&s aren't as bad.
Why, Apple, why?

https://www.dxomark.com/apple-iphone-11-pro-max-camera-review/
Is that stuff seriously supposed to be the best that smartphones have to offer?

[admiralk]

Where this breaks down is that increasing the sensor resolution decreases the pixel size and aperture compromising performance so the result is not quite the same, but of course it is almost always acceptable.  More relevant may be that the existing optics are almost always limiting performance of the lower resolution sensor so using a fraction of the higher resolution sensor image increases the various optical problems.  On close examination, this is readily apparent on my point-and-shoot Canon Powershot where visible coma and chromatic aberration become evident with high contrast sources.

Marketing can make good use of an objective but largely meaningless specification like number of pixels while quantifying something which is more important, like the various forms of image distortion produced by the optics, is difficult.  Sensor resolution passed the bounds of diminishing returns long ago.

As I understand it, if I take a .raw file and straight away save it as a .jpg I would end up exactly what I would get if I save it as a .jpg to begin with. I assume the algorithm used by the software on my computer is more efficient than what the camera uses, so it could even be better, although I doubt it would be noticeable. So the only limiting factors are in the hardware, which is going to be the same in either case.

Diminishing returns are why I only record audio at 24/48. 24/96 doubles the file size for little noticeable difference. As it is, a 2 hour set can easily be over 1 gig per track.

[helius]

As I understand it, if I take a .raw file and straight away save it as a .jpg I would end up exactly what I would get if I save it as a .jpg to begin with. I assume the algorithm used by the software on my computer is more efficient than what the camera uses, so it could even be better, although I doubt it would be noticeable. So the only limiting factors are in the hardware, which is going to be the same in either case.

The RAW converter program has a large impact on the quality. It's true that software on a computer can do more processing, but the camera's image processor is specialized to the task and also (if you use OEM lenses) may have better correction of lens optical distortions.

[EEVblog]

Just get the cheapest used macro lens you can find, and then the cheapest used matching camera.
Results will be way better for a given light then any phone on the market.
The lower the light, the way betterer it gets.

[magic]

As I understand it, if I take a .raw file and straight away save it as a .jpg I would end up exactly what I would get if I save it as a .jpg to begin with. I assume the algorithm used by the software on my computer is more efficient than what the camera uses, so it could even be better, although I doubt it would be noticeable. So the only limiting factors are in the hardware, which is going to be the same in either case.

Maybe things would be different if you use the camera vendor's raw software (if available), but having played with two different 3rd party raw processor I can tell you straight away that in many ways, the results will depend more on the raw software used than on the camera.

There is a lot of processing and secret sauce going into generation of a viewable image from the raw sensor data. For starters, as I hinted before, the uncompressed image you get from this process is actually up to 3x larger than the raw data, go figure Raw has surprisingly little to do with the final image.

Almost every image you see on the Internet, including professional photos, is JPEG. If you have ever seen any good looking images on the Internet, you have seen good looking JPEGs. Raw's utility (if any) is not as much in avoiding JPEG artifacts, as in gaining more control over things like color rendition and white balance, tonality, sharpening or noise reduction. Or outright processing the image beyond recognition for some supposed artistic value. Occasionally, rescuing images which had been destroyed by bad camera settings during shooting.

If you start reading on the topic, you will find rants by professionals who bluntly say they have no time to mess with that stuff and would rather buy a camera which can be tweaked to produce the images they want right off the bat

[David Hess]

Seems like the COVIDAPP doesn't work on iPhones unless the app has focus and the screen is on! 
Useless!

For power and privacy reasons, Apple does not allow background processes full access.

[magic]

Such a shame, the covid app would take the best pictures of PCBs in the world

[admiralk]

Good hijack Dave. 

 @magic I am just guessing here, I have zero experience with anything more than point and shoot.I used to mess around with 3D stuff, but that was mostly modeling, not much rendering. That is about as close as it gets. Even if I do nothing, it is nice to have the option. I picked up the XTi today and it saves both at the same time, so that is nice too.

The camera looks good. Cannot say much else about it until I learn to use it. I did manage to take a crappy picture of my bird though.

Hopefully during the week I can figure out what most of the settings do and how to get to them.

[Electro Detective]

Such a shame, the covid app would take the best pictures of PCBs in the world

It's always the way isn't it ? 

yeah, the pictures would be great for troubleshooting legit bugs 

and halt the spread of the problem at the workbench       

and billions of people won't have to stress about it       

 


----------------

@ admiralk  awesome bird mate     a bit blurred, use flash next time
or tell bird 'hey, stay still!' 

[tooki]

Ironically, phones could have an edge over point and shoots because they use prime lenses which are easier to optimize and make with wide aperture. I mean, it is an edge as long as you are OK with whatever angle of view the manufacturer chose for you But anyway, it seems that about f/2 is the norm for phones these days, which enables a few megapixels worth of detail on a 1/3"~1/2.5" sensor if aberrations are taken care of.

That being said, I stumbled upon a review of some super-duper iPhone 11 Pro Max today, apparently Apple's TOTL phone, and I am absolutely not impressed by their sensor / image processing performance. A bunch of sample photos shot in bright sunlight and every single one has painful noise reduction artifacts all over the place. Somehow small sensor p&s aren't as bad.
Why, Apple, why?

https://www.dxomark.com/apple-iphone-11-pro-max-camera-review/
Is that stuff seriously supposed to be the best that smartphones have to offer?

I pored over that review and cannot find those “bright sunlight” pics with “noise reduction artifacts all over the place”. In the low light shots, yes. In the bright sunlight, absolutely not.

[magic]

Smudges on water, posterization on clouds, thick grain on other clouds and elsewhere on water. Some of the grass is suspicious too but I'm not entirely sure, the underlying real texture of grass overpowers the crap. Just the general look of overzealous noise reduction that I can't stand

For comparison, similar scene from a "real" camera with similar size CMOS sensor and similar pixel density. It's not stellar and there is some noise too, but at least it goes away when you scale to 50% (which is a much more reasonable megapixel output from such tiny sensor anyway). The phone's artifacts don't disappear at 50%.
https://www.imaging-resource.com/PRODS/canon-sx60/canon-sx60GALLERY.HTM