Why we've reached the end of the camera megapixel race | Hacker News Reader

Why we've reached the end of the camera megapixel race

arstechnica.com

15 pointsevo_932 comments

I'd like to see a shutter-speed race. My last 2 (inexpensive) cameras were slow on the draw and I frequently missed good shots of my kids. If I tried to take an action shot, it was nothing but blur.

that is a 'sensitivity' race, actually...

Shutter speed, ISO, and Aperture in photography are related in much the same way as Voltage, Current, and Resistance.

To change any of those three (in your case shutter speed) you have to compensate in one or both of the others. Lens aperture is limited mainly by form-factor (can't have a large aperture in a tiny zoom lens) and is therefore rather constant across Point-and-Shoot cameras. but Sensitivity can be improved by having higher-quality sensors and supporting circuitry.

My first P&S camera had a maximum ISO of 400 and was very noisy at that level. my first DSLR did 1600 (4x as sensitive) and the current bleeding edge is 128000 (8x more sensitive again). Even in inexpensive point and shoot maximum ISO seems to be around 1600 (or more).

I'd like to see a quieter shutter (and mirror). There are a number of situations where I have to put away the DSLR and shoot with a good point-and-shoot just to avoid disturbing other audience members. I guess I need to try a mirrorless micro 4/3rds.

I'm excited about a possible future of curved sensors, transparent sensors, full-resolution stills from video (especially shooting kids and sports), etc. But a lot of times all I really want added to my DSLR is more software options. I love how my old Canon S3 IS has a "Custom" setting that I can configure. I want about 5 "Custom" settings on my DSLR so I can quickly flip to my own set of preferences - ISO, metering mode, resolution, etc. - depending on the situation.

> "But a lot of times all I really want added to my DSLR is more software options"

I'll second that. A set of user-definable presets, tap to focus, auto-HDR capture ... that would be so great. It's frankly frustrating that Canon, Nikon, et al can be so far behind $1 software on consumer smartphones.

Canon, Nikon, et al are making those cameras for photographers, not smartphone users. The interfaces may be alien to a newcomer but are really just power tools for working photographers.

I'm pretty sure even a "real" photographer would appreciate being able to change the focal point mode in less than 5 button presses through a 80s-era menu interface.

I'm not a "real" photographer but I do appreciate being able to change the focal point mode via a button and a wheel. Fortunately my (serious amateur level) DSLR camera provides this feature via programmable buttons.

I've heard of some alternate firmware for cameras. I even used it to automatically take a picture every 30 seconds when sending the camera on a weather balloon.

I'll point out another reason why you might not want more megapixels - size and data management. If you're shooting your photos RAW and you take a lot of them, you'll easily have a photo library whose size is in the hundreds of GBs, if not in the TBs range. Hard drives are cheap, but in all likelihood your photos are precious to you - meaning you have to back the library up on a regular basis.

A glut of meaningless megapixels makes it harder to preserve the meaningful data buried within.

Speaking of 12MP and low light performance, no mirror and a quiet shutter, I've got a Fuji X100 in the mail, finally after two months.

http://www.finepix-x100.com

Thanks for posting that. I've fantasized about a Leica M9 but the body and the glass would run about $8,500. Looks like a (much) lower priced alternative.

Well if you got the glass cheap enough, but really you'd probably be spending closer to $10k.

These are in and out of stock quickly right now. I preordered two months ago and I'm just now getting it shipped from B&H.

This camera is really the answer to Leica's X1, and most people think it's better than the X1 for various reasons. The general consensus is that it's an awesome camera with some firmware glitches that need to be ironed out. I've been shooting with a Canon S95, and although it's an amazing little camera, I realized just how much you really need a viewfinder. But the Fuji is going to be everything I've really been wanting: a smaller camera with a fast semi-wide lens lens (35mm) with great low light performance.

Megapixels haven't meant anything to me in a decade. I'm more concerned about shutter speeds, lens, and image compression.

Disclaimer: I am a point & clicker. I've never even touched a DSLR camera.

The marketing idea that consumers could judge a camera based on 'how many megas it has' died a while back. It was definitely dead in 2009 when Sony released cameras with Exmor backlit sensors and competed with a similar Fuji offering.

In the professional market, however, it's possible to get an 80 megapixel medium format sensor and there are post-production advantages that make this worthwhile for professional use. I wouldn't be surprised if sensor resolution continued to creep up in prosumer cameras as well.

There are still benefits for consumers from higher sensor density and resolution, as long as it can be done without introducing noise. For example, slightly less awful digital zoom and the ability to liberally crop.

You'd think people would avoid saying that 'X should be enough for anybody' after the whole Gates thing. (Yeah, I've heard it's a misquote, etc... Doesn't matter, it's famous and should serve as a warning anyhow.)

Not really. I'm prepared to go out on several limbs here.

"Four wheels ought to be enough for anybody"

"Twelve beers ought to be enough for anybody"

"Two intestines ought to be enough for anybody"

The point being that beyond a certain point more pixels (unlike more memory) just makes life worse. There's only so many photons to go around.

> "Four wheels ought to be enough for anybody"

Most truckers now agree that 18 wheels is the best. That's after number-of-wheels race from 70s where we tried upto 22 wheels.

Or perhaps 10 is 2.5% more efficient and saves money up front as well. Go evolution!

http://www.epa.gov/smartwaytransport/transport/documents/tec...

Sigh. I knew someone would reply to that with a comment about trucks.

edit: Fun fact: most truckers also agree that fourteen beers is better than twelve.

Sort of agree but with image sensors it's a bit different: even if your process allows you to jam more and more pixels onto the sensor the number of photons landing on the entire sensor doesn't go up unless you're willing to schlepp a larger lens around. Also, the smallest printable dot on photo paper hasn't improved that much in recent years.

I'd take a 6 MP hit (down from 12 MP to 6 MP) for an extra two stops in speed. I have a D90 with an 18-200mm VR lens. It's tremendously versitile, and the VR helps shooth handheld at long focal lenghts in less than favorable light, but indoors, I either have to crank up the ISO or mount a prime.

Newer DSLRs have even better low light performance than my camera, but at ISO 800, the noise and artifacts begin to bother me. I'd happily trade a few MP to get ISO 200 image quality at ISO 800.

I recently printed some 15 mega-pixel images in 36"x24" and they quality was quite good. The number of people in the world who need > 24 mega-pixel images is very very small.

Only [5%](http://www.canonrumors.com/forum/index.php/topic,818.0.html) of photographers want a mega-pixel bump in the new 5D MK III.

Last summer, when I got one of the then best Android smart phones on the market, from all the cool features and interesting capabilities, the question I got asked the most was: "How many megapixels does it have?".

It took me some time to get used to it, and not get annoyed every time. People like to focus on a single number for quality measurement.

I hope not - while megapixels alone aren't sufficient to produce a great photo, they are a necessary component. I don't mind if the industry pauses for a while to increase the image quality through other means, but eventually the pixel count will become the bottleneck again if it's not increased.

I'm still using an ancient 12 megapixel camera, and I have yet to encounter a situation where I was unable to produce a great photo due to an insufficient megapixel count.

First you need to resolve the other bottlenecks or more megapixels will only make the inadequacy of the lenses more obvious.

The lesson to learn here is that the camera needs to grow more organically, it's useless to grow megapixels so much without improving lenses or the size of the sensor, instead, as the article states, it makes it _worse_. Like megahertz, megapixel is a nice single number that many customers associated with performance but it's pretty far from truth. A more detailed sensor will only see what the lenses let go through and a smaller sensor will become that much less sensitive. It is cheaper however.

There is no such thing as 'too many pixels'.

Case in point: http://skysurvey.org/

Actually there is, fundamentally. When you start having more pixels than your system can resolve (because of motion blur, lens defects, etc...), then you have too many pixels.

I think you have too many pixels if you can't get low noise performance at ISO 800 equivalent.

But there is such a thing as 'too many pixels per mm'.

Reply on news.ycombinator.com