Report: Apple's next iPhone to sport 3.2-megapixel camera - Page 4

Originally Posted by gmcalpin

No shit? That's not the point that I was addressing, and you know it. :P

Also, IMAX isn't shot digitally; it's shot on 70mm film.

Originally Posted by iReality85

So maybe I'm just missing something? You say that more MPs are pretty much proven to be useless (and therefore do not add to the quality of an image), but then you go on to correctly say that for consumer devices like "point-and-shoot" cameras the additional MPs compensate for the lack of superior optics. I understand how this operates and have understood that all along. But for arguments sake, don't those additional MPs increase the quality of images, merely in a round-about way due to lesser grade parts and technology? Sure, industrial HD cameras are able to produce HD images at a mere 2MPs, but you're also talking thousands of dollars for such capability, not a consumer grade "point-and-shoot" device. I guess my point has been all along that, sure, the iPhone is capable of outputting HD imagery, but I would only rank its current 2MP as "satisfactory" due to the fact the iPhone does not sport any fancy optical equipment, etc etc. Therefore, given this, wouldn't an increase in MP to 3.2 (or anything above) come as an added benefit and increase the overall image quality?

Originally Posted by quinney

He doesn't want to have a discussion. He wants to push people's buttons and provoke
a reaction. Notice how often his posts include references to fanboys, drinking koolaid,
and Apple products sucking. I was given some hope last night when I read a post from
Solipsism, where he seemed to be getting tired of bickering with Teckstud. Now you seem
to be volunteering to step in and satisfy Teckstud's neediness. Spare us please and just
ignore him.

Originally Posted by digitalclips

Sorry, I meant to say RED cameras not IMAX and mentally blinked ...

Originally Posted by melgross

Well, perhaps, sorta.

The problem with these devices is that you can't get RID of the depth of field!

That makes good photography impossible.

Originally Posted by Virgil-TB2

What I don't get about the moderation here is how people like "teckstud" never seem to suffer from their constant personal attack posts. I try to be civil as much as possible, but when I veer even slightly off course I get a warning from the mods.

I recently reported on five posts in a single thread by "teckstud" however, where he was just calling people all kinds of names and going on in the rudest manner. He was being personally insulting to about a half dozen people which is a total violation, yet later that same day he is posting as usual.

He seems to have a few alts as well, so banning him for a day or two is hardly likely to cause any problems for him anyway.

Maybe he's related to Kasper? Or perhaps he is just a "character" to make the forum more lively and is actually run by one of the mods.

Originally Posted by gmcalpin

Well, now you're talking. I would sell my niece for one of those.

Originally Posted by digitalclips

Is this this theory based on; light is a wave, a particle or is this quantum mechanics?

I am being facetious, sorry. I just meant show me a camera in the same light conditions with a larger MP sensor (with associated lens) that can't do a better job.

Originally Posted by digitalclips

Sorry, I meant to say RED cameras not IMAX and mentally blinked ...

The Red info is a good indicator of sensors quality, drool. I quote ..

"Typical high-end HD camcorders have 2.1M pixel sensors and record with 3:1:1 color sub-sampled video at up to 30fps. RED offers the Mysterium ™ Super 35mm cine sized (24.4×13.7mm) sensor, which provides 4K (up to 30 fps), 3K (up to 60 fps) and 2K (up to 120 fps) capture, and all this with wide dynamic range and color space in 12 bit native RAW. At 4K, that’s more than 5 times the amount of information available every second and a vastly superior recording quality. In addition, you get the same breathtaking Depth of Field and selective focus as found in film cameras using equivalent 35mm P/L mount lenses. Mysterium ™ boasts greater than 66db Dynamic Range thanks to its large 29 sq. micron pixels. And 12,065,000 pixels deliver resolution that can only be called Ultra High Definition."

Originally Posted by melgross

First of all, we have to talk about the same generation of devices. It's not (usually) possible to discuss different generations, with different resolutions, and come up with a useful answer.

I said much earlier here, that in a few years, Apple could likely put a 5 MP sensor in, and have a better picture than a 3.2 now.

We are making progress. If we look at spanning generations of product, we will see that higher resolution cameras today, with the same physical size sensor than those of earlier generations have better IQ overall.

But, when you look at cameras from the same generation, you will see that cameras with a substantially higher rez chip (assuming the cameras are priced about the same), or, a chip with about the same rez, but a smaller physical size (APS-C vs 4/3, or ff 35mm) will have worse performance overall.

It comes down to pixel site size. The smaller the sensing site is, the less light it takes in, and the greater the percentage of signal the dark current from the site affects the areas of the image with less light. The brightest areas may be unaffected, but as it gets down to the mid tones, noise intrudes. Dynamic range is also lessened.

As the ISO is increased, noise moves up into the lighter parts of the image until the entire image is affected, and the darker parts are unusable. Detail also disappears at the higher ISOs, because the "grain" begins to obscure it.

Then we have noise reduction. The poorer s/n the sensor has, the more noise reduction is needed, which also obscures detail.

What happens is that a camera with a high rez according to specs, can often have less rez at higher ISOs than a camera with a substantially lower rez.

These problems can pop in at very low ISOs if the sensor is small, and the rez high.

For example, the Canon G10, a "pro" compact has a very high IQ at ISO 100, its base ISO. But when it gets to 800, it's about useless.

This camera has 14.7 MP, and a good lens.

The Canon 50D, a D-SLR, has 15.1 MP, but the image IQ regarding s/n and dynamic range is much better at 3200, and even 6400.

But that camera's s/n is slightly worse at lower ISO's than the older 40D which is a 12 MP camera. It is better at higher ISO's though.

How far do you want to take this? It's a very complex subject.

Originally Posted by Virgil-TB2

What I don't get about the moderation here is how people like "teckstud" never seem to suffer from their constant personal attack posts. I try to be civil as much as possible, but when I veer even slightly off course I get a warning from the mods.

I recently reported on five posts in a single thread by "teckstud" however, where he was just calling people all kinds of names and going on in the rudest manner. He was being personally insulting to about a half dozen people which is a total violation, yet later that same day he is posting as usual.

He seems to have a few alts as well, so banning him for a day or two is hardly likely to cause any problems for him anyway.

Maybe he's related to Kasper? Or perhaps he is just a "character" to make the forum more lively and is actually run by one of the mods.

Originally Posted by melgross

You've looked at the price of that body (which is considered cheap by motion picture standards), and lenses?

It's a bit out of the iPhone range.

Originally Posted by melgross

First of all, we have to talk about the same generation of devices. It's not (usually) possible to discuss different generations, with different resolutions, and come up with a useful answer.

I said much earlier here, that in a few years, Apple could likely put a 5 MP sensor in, and have a better picture than a 3.2 now.

We are making progress. If we look at spanning generations of product, we will see that higher resolution cameras today, with the same physical size sensor than those of earlier generations have better IQ overall.

But, when you look at cameras from the same generation, you will see that cameras with a substantially higher rez chip (assuming the cameras are priced about the same), or, a chip with about the same rez, but a smaller physical size (APS-C vs 4/3, or ff 35mm) will have worse performance overall.

It comes down to pixel site size. The smaller the sensing site is, the less light it takes in, and the greater the percentage of signal the dark current from the site affects the areas of the image with less light. The brightest areas may be unaffected, but as it gets down to the mid tones, noise intrudes. Dynamic range is also lessened.

As the ISO is increased, noise moves up into the lighter parts of the image until the entire image is affected, and the darker parts are unusable. Detail also disappears at the higher ISOs, because the "grain" begins to obscure it.

Then we have noise reduction. The poorer s/n the sensor has, the more noise reduction is needed, which also obscures detail.

What happens is that a camera with a high rez according to specs, can often have less rez at higher ISOs than a camera with a substantially lower rez.

These problems can pop in at very low ISOs if the sensor is small, and the rez high.

For example, the Canon G10, a "pro" compact has a very high IQ at ISO 100, its base ISO. But when it gets to 800, it's about useless.

This camera has 14.7 MP, and a good lens.

The Canon 50D, a D-SLR, has 15.1 MP, but the image IQ regarding s/n and dynamic range is much better at 3200, and even 6400.

But that camera's s/n is slightly worse at lower ISO's than the older 40D which is a 12 MP camera. It is better at higher ISO's though.

How far do you want to take this? It's a very complex subject.

Originally Posted by digitalclips

partial quote:

Yep I agree, they bring pin-hole cameras to mind ... wait a a minute ... they are pin hole cameras!

Originally Posted by digitalclips

I was obviously assuming that any two cameras being compared in my question were of the same technological moment in time, why on earth would you assume otherwise?

I worked for a decade with Scitex, Leaf and other high end scanner companies so could bore the rest of the readers with technical blurb too, it doesn't get away from my basic premise is that at a given price / performance level (yes same era) more is usually better all things being equal. Look at the specs on the RED video camera I posted! Granted one manufacturer could obviously stick in a larger chip with lousy optics hence the generalization of assuming all things are equal.

Originally Posted by digitalclips

Can't argue there lol.

I dream of it most nights.

The point of mentioning it is that I always try to argue with the 'We don't need more' brigade.

We would never need more than 1MB RAM and a 100 MB hard drive etc ...

One day ... not too far away the iPhone (or its descendant) will surpass the RED of today ... and be 3D

Originally Posted by melgross

Of course, and they are very good for those special photos that a few people take from time to time.

But would you seriously want every image to look like that?

Would you like to take a portrait and have every stitch of the background cloth be as sharp as the subjects eyelashes?

This is what a diaphragm is for (No, not THAT kind of diaphragm!)

We all want fast lenses so that we can shoot in darker venues, and so that we can open up to get that razor thin depth of field when we need it to separate the subject from the background, as one example.

If we want the depth, we can close down.

These tiny sensor cameras don't allow that at all. Every picture must have depth from almost one foot to infinity. They'll never put a fast lens in, because that would cost too much, and be too big.

Originally Posted by teckstud

Apparently to infinity as long as you're discussing it. Talk about making mountains out of mole hills. We are discussing adding to the iPhone a 3.2 MP camera. Why don't you talk to us about the Hubble telescope camera while you're at it?

Originally Posted by digitalclips

All true BUT you assume some of the analog processes of which you speak could not be performed digitally. I can imagine all of the depth of field controls currently done with aperture / speed variations being controlled with a few sliders on a touch screen.

Originally Posted by melgross

Eventually, that might be possible. But it would require much more intelligence on the part of the program. How would it recognize what part of the image should be in focus, and by how much. How would that slider "know" what to affect? Would we have to make image selections first? How accurate would that selection need to be. What about depth inside the selection. We don't want that artificial cardboard cutout 3D effect.

We can do this in a way now, by taking photographs with different focus, and using a program that combines them in CS4 to make one image. It's crude, but it can work in a post way. But it's not really accurate. You can't pick what you want to be in focus, and what you don't want to be in focus. There are other problems too, but it's sometimes ok when you need the entire image to be in focus.

Originally Posted by digitalclips

Well leave PS4 out of this were talking about was depth of focus not post production! Alll that is required is the speed /ISO/ and aperture settings to be manipulated. It would be simple to have sliders for advanced users and also an 'idiots' option with various sample images demonstrating different results from certain presets. This already exists in many low end cameras anyway so i am not suggesting anything new, merely that Apple could make a really cool set of controls with the SDK. Think about it, they could put SLR type control on the screen!

Originally Posted by melgross

I can't see how that would work. So you focus on a subject, but not doing it optically, where you would normally close the lens down, or open it up, it would be done in software. I think that's what you're suggesting.

I have nothing against the concept. It would be brilliant if it could be done.

ISO settings are already done in software, not with a slider, but in software all the same. But the actual work is done by amplifying the signal.

Let's get to the level below the onscreen controls. How would the camera actually do this in software?

You said that I'm thinking in an analog way. i guess I am, as the sensors are analog, and the ISO changes are done in analog. The only thing there that is done digitally is noise reduction, and even there, before the processing, which turns it into a digital signal after the amplification is done, and where most of the NR is done, there is analog NR.

So, what are you saying? What I thought you said at the top of my post, or something else?

Originally Posted by Lictor

If Apple wants to improve the quality of the photos from the iPhone, the recipe is known, and it doesn't involve increasing the pixel count.

What is the purpose of the camera in a phone?
Is it to output A4 (letter) size? FYI, a good quality 3mp camera is enough to output A4 print, I have been doing it for years with my Coolpix 990.
...
So, unless you intend to print bigger than A3 or to be able to zoom inside the photo on a 24" monitor, 2mp is already enough... If this is the target, 3mp is nice, because it allows for some room to crop, but beyond that, you don't gain much.

The human eye, under best lighting conditions, can resolve objects slightly less than one arc minute (Blackwell, 1946; reference and plots are in my book Visual Astronomy of the Deep Sky, click here to go to my visual astronomy pages ). One arc-minute corresponds to 0.003 inch at a distance of 10 inches. The inverse of 0.003 = 344, but it takes at least two pixels to resolve something, so double this number and we get about 700 ppi as the resolution of the eye at one arc-minute. The eye can do a little better, so perhaps 1000 ppi is the limit.

Originally Posted by melgross

Why don't you actually do some good, and give us a definitive argument about this, instead of making useless comments?

Everything I said, as most people likely realize, applies to low rez cameras with tiny sensors as well, as was the point. Most people are likely also familiar with the compact and D-SLR camera rez vs IQ argument.

Come on. Give us some of your understanding of what should be done and why. I promise to read it. Really!

Originally Posted by digitalclips

I may have drifted off your post, can't remember ... I am editing in FCPro for a client on two screens and following you guys on a third ... So forgive me if I lose track

I realize the lack of optics means it is guess work, so the idiot settings would work fine as in CLICK HERE FOR 'close person background blurry' setting etc. For me I'd rather be able to change settings in a 'manual' way and have a visual light meter perhaps showing over areas I can select. I can throw open the shutter and crank the speed to hold the reading on the desired area knowing the depth of focus just dropped even if I can't see it.

Originally Posted by teckstud

What should be done is what's being done- upgrading the camera and adding video capability to the iPhone itself. I have been arguing that these were features that I would like on the iPhone for months - which you and your friends like to label as "complaining". Obviously I was right and now you're try to explain it to us all.
There is nothing more that needs to be done except read you dissertations now on how light is refracted , etc, etc, etc. I'm looking forward to your biography of George Eastman next.
And I can't wait for future threads where you explain the quantum physics of MMS texting and pics.

Originally Posted by melgross

That's ok, one third of your brain is better than two thirds of some others.



I think we're talking past each other, which is why we're not getting together on this. That's why posting isn't always the best method of communication.

Canon used to have, in the '90's, a system where it would track your eye movements for focus. They dropped it when they came out with digital. Too bad, as it worked pretty well, and with advances, it would work much better today.

The system could be expanded so that you could look off frame, say, to select a function which would then appear in frame, and allow you to adjust it with your eye. This could work. But it would still be making the actual adjustments the old fashioned way.

Originally Posted by melgross

Your response is reasonable. I never disliked the idea of video on the phone. Though I did say that most people didn't seem to think they needed it.

But, if you stopped being such a wiseguy, then people would stop complaining about you, publicly, and to the mods.

Just respond without the wisecracks.

All of us have been talking to one another without problems. We don't all agree on everything, we don't always understand what others are saying at first, but the only one who is having problems here is you. Why is that?

And, no, it's not me.

Originally Posted by digitalclips

Ha ha ... can I quote that to my wife ?

I am probably describing something more simple - NOT. OK let me describe what I am seeing in my mind's eye: This assumes a pretty nice lens from apple with a wee motor not a pin hole fixed and a minute focusing system. OK OK I am dreaming

Then assume we have a full screen image on iPhone as I pan around to compose. First I set the ISO manually say from a range of 400 to 1600 (this will depend on chip) for the desired results. Now I can draw with my finger an area which is used as focusing point and also gives me a light meter level (-3 -2 -1 0 +1 +2 +3 ) with a floating needle indicator and also I have two graphical controls for aperture speed and shutter speed I can alter.

Voila I may not see depth of focus changes but I am experienced enough to get a pretty good result with the afore mentioned controls.

Next I need the beam me up option ...

Originally Posted by melgross

Ok, its coming into focus (sigh!) now. so you're not opposed to the work being done in an analog way, just having on screen controls?

Ok, first of all, while that could work, you would really want to just touch a spot for focus. Possibly even magnify that spot (perhaps it could call up magnification automatically!) so you could pinpoint the very point you want in best focus.

Maybe you COULD circle an area that would be used for the metering, and then it would find the center of that spot for focus, with the magnification so you could adjust it.

I think that could be done faster and better with an eye controlled system. Otherwise, this would take a while to do, and you would have to hold the camera with one hand, away from your face, make the adjustments while making sure nothing moved, then taking the picture quickly, with one hand, as you wouldn't have the time to get to a two handed grip, unless the subject was static. If you move the camera just a bit during this, every adjustment could go off.

ISO of 400 to 1600. Ho ho. That's pretty high for a flyspeck sized sensor. Maybe 40 to 200.

Originally Posted by digitalclips

That is an over simplification. The excellent web page article you linked to points out that the square law means the difference of "40% in the pixel count in one direction after doubling the total pixel count". To then interpret that to mean 'More MP is proven useless" is a silly extrapolation. If you were correct in this assertion then all cameras need only be 1 MP!

I agree the hype of manufacturers is over done at the consumer level and the % difference between any two cameras should be calculated before paying a premium for a few extra MP. It is simple arithmetic after all.

However, more MPs are better than less MPs if the price and storage are not issues and lots more even better. Anyone who works a lot in Photoshop knows the benefit in working in a high resolution for such things as masking and color replacement even if the final image is to be used at a lower resolution. The accuracy of the editing is significantly better when done at the highest resolution the graphic artist can comfortably work at for his own hardware's abilities.

There is a well known truism in graphics and video, 'you can always res down but you can never res up'.

Originally Posted by Máedóc

Why is nobody here discussing the very real possibility that the camera Apple has chosen could be using the Omnivision Truefocus technology as mentioned by retroneo in the 8th post on this thread?

Truefocus (aka: Wavefront Coding) is a paradigm shift in the history of imaging and optics. Originialy developed by the Imaging Systems Laboratory at the University of Colorado, the technology was privatized by CDM Optics in order to develop products for the scientific and medical industry. CDM was aquired by Omnivision in 2005 opening the door for consumer application of Wavefront Coding technology.

From the Omnivision website:
“A revolutionary technology called Wavefront Coding™ is set to fundamentally change the nature of digital imaging. Wavefront Coding is the joint optimization of non-traditional, specialized optics and sophisticated signal processing. Unlike conventional imaging systems, with Wavefront Coded systems optics and signal processing are closely tied together to facilitate a new and improved image formation process that produces images with significantly increased depth of field without sacrificing light gathering. This close relationship between optics (encoding) and image processing (decoding) enables digital cameras to capture images that are sharp and clear at any time and throughout the object field without the need to physically move optics to focus. Wavefront Coding enables true 'point-and-shoot' digital cameras. Because Wavefront Coded systems shift a large portion of traditional optics and the auto focus mechanics into novel fixed optics and silicon, Wavefront Coded camera systems enable reduction in overall camera module size as well as continued systems cost reductions that follow Moore's Law.

Wavefront Coding is now for the first time available in commercial applications, more specifically mobile-phone cameras where proximity focusing, low height optics and low tolerance assembly are combined with high target yields. Marketed under the name TrueFocus™, Wavefront Coding is used to optimize optics, sensors,and processors to maximize imaging performance and minimize costs. Such advances have not been possible until now. Using TrueFocus technology offers real and instant one-click photography without delay in image capture or spoiled pictures for the end user, effectively paving the way for the next wave of affordable camera phones offering superior imaging capability at higher resolutions without the need for auto focus. “

Although Omnivision also makes traditional image sensors, I think it's highly likely that Apple will use their TrueFocus™. If thats the case then all this talk of depth of field, sensor size, number & quality of lens elements, etc. will need to be reconsidered.


http://www.ovt.com/products/truefocus.php

http://www.cdm-optics.com/?section=Tutorials

Originally Posted by m2002brian

Said "...pretty much useless" in the case of the iphone going up from 3.2 to anything any other camera phone would have.

Originally Posted by Virgil-TB2

What I don't get about the moderation here is how people like "teckstud" never seem to suffer from their constant personal attack posts. I try to be civil as much as possible, but when I veer even slightly off course I get a warning from the mods.

Originally Posted by teckstud

Well I am mod.
But you fail to realize that I usually get attacked first , like you just have,with usuallly something entirely not related to my posts, which are always topic related.

Separately, DigiTimes claims that OmniVision "is also said to have secured 5-megapixel CIS orders for another Apple product expected to be launched later in the year." The fabless CMOS image sensor designer reportedly beat out STMicroelectronics and Aptina Imaging for the orders.