EVO camera blind test

[reemas]

cool info spork

 

[mrspeedmaster]

Now, look at the iPhone 4 videos just leaked from the Czech republic. These are real videos. I checked the data of the file and downloaded the source movie.

Score! First Real World Apple iPhone 4 Photos / Video | Obama Pacman

Watch the original file and look at the blade of grass. Impressive.

 

[Aldo Junior Ao2]

I gave you a thank you because you helped me prove a point. Case closed have a goodnight.

 

[SporkLover]

View attachment 6591

I have the lights off with the Tv paused. This is what 9 frames per second looks like buddy. So When you made that video you had the lights off like me with only some parts of the tv screen to light you up. Shoot i even dirty the lens just to get it that low. Come on guy you were caught. Be honest 9 frames per second?????? You must have a lemon then.

I'm not seeing where you proved a point.

Less than optimal lighting and the camera sensor trying to compensate with metering adjustments is enough to kill frame rates.

 

[dr g]

Brush up on your Digital Camera language folks. The Digital sensor in the EVO is a newer generation of sensor than whats in the Iphone 3GS, and it's manufactured by the same company, same technology, same production processes..... the words cheap is a poor way to describe the sensor. More Mega Pixel does not equal better quality. Generically, you getter better quality from larger pixels (pixel pitch) and larger camera sensors. In reality there are many specs to the camera sensor that really dictate the "quality". I'll try to keep it as simple as possible. The Iphone 3GS has a 1/4 camera sensor, the Evo a 1/3.2 camera sensor (slightly larger). The Iphone's sensor is divided into 3.1M pixels providing a 1.75 pixel pitch (Large), the Evo's sensor is divided into 8M providing a 1.4 pixel pitch (small). What this means is that although the Evo has a larger sensor and more pixels, those pixels are less "sensitive" because they have less area to capture light. This correlates directly to why low light conditions would lag in the Evo. The Iphone has larger pixels, better capable at capturing more light per pixel. In good lighting the Evo should have superior pics. Since this same sensor is responsible for video, the same applies to video capture for each phone. Now that was speaking from a purely hardware specs comparison of each sensor. The design of the sensors give the manufacturers of the phone a great deal of freedom over the operations of the sensors (metering, etc etc etc).... so the other part of that equations is going to be the software written to control the camera sensors. If I were to describe the sensor in the Evo, I would call it a marketing trick that plays into the consumers line of logic that more MP = better pictures. The engineering behind it produces 8MP, but it hampers sensitivity of the sensor, which can introduce the blurring you guys see. This is very reminiscent of the Intel vs AMD clock speed wars, where AMC proved with the Athlon line of processors that you can have a faster processor with lower clock speeds. Just to set up some of the fragile egos for the Iphone 4 release..... it will have the same size sensor as the Evo, with 5MP, giving it 1.75 pixel pitch..... theoretically producing better pics!

... on a per-pixel basis. That's the rub, particularly when downsampling, the individual pixel pitch is only directly comparable if displayed at the same size, pixel per pixel. As in digital cameras, a higher resolution often hides the individual pixel noise better when displayed at normal sizes.

This means that the evo sensor is theoretically capable of as good or better low light imaging at VGA resolution. It's a question of the software and, perhaps, the associated hardware (buffers, etc).

What we are likely seeing is the effect of a lack of time and money invested in the imaging software and/or camera firmware.

One potentially useful feature is the flash, which can be used as a video fill light.

 

[jonathan76888]

hahhahahah, Cmon No iphone would win in Phandroid, us Android fans are more loyal than apple fans!! so w.e you say you are wrong!!
the Evo Rules, Android Rules, And apple and Iphone, could go to hahahhaha

 

[SporkLover]

... on a per-pixel basis. That's the rub, particularly when downsampling, the individual pixel pitch is only directly comparable if displayed at the same size, pixel per pixel. As in digital cameras, a higher resolution often hides the individual pixel noise better when displayed at normal sizes.

This means that the evo sensor is theoretically capable of as good or better low light imaging at VGA resolution. It's a question of the software and, perhaps, the associated hardware (buffers, etc).

What we are likely seeing is the effect of a lack of time and money invested in the imaging software and/or camera firmware.

One potentially useful feature is the flash, which can be used as a video fill light.

Well in general I would agree with that, because Larger sensors would mean more pixels, and larger pixels. Larger and more pixels can definitely hide noise. The Evo doesn't fit that though, the pixels are smaller. Smaller and more tightly packed pixels produce more noise and have less dynamic range, all of which contribute to mediocre to disappointing low light performance.

I agree on your 2nd and 3rd points, I'm sure better performance can be had with software tweaks, and the flash is a must for low light conditions.

 

[dr g]

Well in general I would agree with that, because Larger sensors would mean more pixels, and larger pixels. Larger and more pixels can definitely hide noise. The Evo doesn't fit that though, the pixels are smaller. Smaller and more tightly packed pixels produce more noise and have less dynamic range, all of which contribute to mediocre to disappointing low light performance.

I agree on your 2nd and 3rd points, I'm sure better performance can be had with software tweaks, and the flash is a must for low light conditions.

Dynamic range is an issue but noise is not, noise is a per-pixel issue and having more of them simply makes it less of a concern. There are plenty of examples in the digital camera world of smaller, denser pixels performing as well or better than larger, less dense pixels on a sensor of the same size. A lot has to do with software, the quality of the sensor itself and, something we have not yet even started discussing, the lens in front of the sensor.

That said, the iPhones do seem to pump the bitrate out, which leaves me wondering if there is a hardware limitation somewhere on the HTC devices. Perhaps a bandwidth difference between the SD vs inbuilt storage interfaces?

Evo videos tickle along 30fps in the best conditions (and lower resolutions) while the iPhone4 video is rock solid 30 at full 720p. Why the framerate differential?

And perhaps more importantly, why does HTC continually give camera performance short shrift? Does it just not have the resources to fully exploit the hardware?

 

[SporkLover]

Dynamic range is an issue but noise is not, noise is a per-pixel issue and having more of them simply makes it less of a concern. There are plenty of examples in the digital camera world of smaller, denser pixels performing as well or better than larger, less dense pixels on a sensor of the same size. A lot has to do with software, the quality of the sensor itself and, something we have not yet even started discussing, the lens in front of the sensor.

While there are great examples of small sensors performance out of compact point and shoots where high quality sensors and optics can address the issue of the noise introduced when lots of pixels are packed onto a small sensor, I would submit that in the mobile phone world, those same measures haven't been implemented, otherwise we'd have a much more expensive phone on our hands. That being said, you get a mediocre sensor like the ones in the phones, you still get noise issues from more pixels on a small sensor. Granted, as you've pointed out, that more pixels are much better at hiding noise.

That said, the iPhones do seem to pump the bitrate out, which leaves me wondering if there is a hardware limitation somewhere on the HTC devices. Perhaps a bandwidth difference between the SD vs inbuilt storage interfaces?

Evo videos tickle along 30fps in the best conditions (and lower resolutions) while the iPhone4 video is rock solid 30 at full 720p. Why the framerate differential?

And perhaps more importantly, why does HTC continually give camera performance short shrift? Does it just not have the resources to fully exploit the hardware?

I've read on other forums that part of HTC's issues with other phone models was they did not pay the licensing fees for the appropriate drivers to "unlock" full potential of the camera sensors. The Touch Pro and Touch Pro 2 models immediately spring to mind. Those phones had paltry camera performance when it should have been better.

I've never seen proof.... but interesting theory that feeds the Software angle.

 

[dr g]

Makes you wonder if they shouldn't slow down on the different models.

 

[HarryKeogh]

All this techno-talk just put me to sleep. Look, I want to be able to grab my cell phone and take video of my niece eating ice cream while she gets half of it on her face and have the video look half-way decent. The EVO video was a joke. HTC needs to step it up.

 

[JunBringer]

Invalid or not, just look at the video and vote with your eyes. I don't think anyone cares about sensor jargon. Period. There is a reason HTC is advertising the 8MP camera. They know people want "more" megapixels regardless of quality.

And I have a good camera, but I also have other nice things.

The point is not what I have, but how HTC's 2010 Camera compares to the competitions 2008 camera.

I thought the 3GS was 2009?

Either way thanks for the comparison and I'll happily admit to anybody that iPhone has a better camera, and of course better games and a few other things. Like somebody recently posted, you gotta give credit where credit is due. Both are awesome phones and have their ups and downs. I loved my iPhone and I love my EVO.

 

[OfTheDamned]

There isn't any reason to call someone an Apple fanboy or troll just because they post something that you don't like. Everyone praised the OP for doing the test until he revealed which phone was which and suddenly he is an evil fanboy? The rudeness is simply uncalled for. If you don't like his results, then conduct your own test. There are enough iPhones in the world that I'm sure you can find someone to help you if you don't own one already.

Please keep in mind that we do have a zero tolerance policy and it does apply to everyone. If you can't be polite then don't post. If the insults start to fly again, infractions will be handed out.

 

[EarlyMon]

Nice test, and I appreciate the effort.

At first, I found the Evo to be terrible for still photos, despite a great thread of great snaps to the contrary. I then decided to turn off auto, set a constant ISO (typically 200, sometimes 100 or 400), definitely check White Balance and adjust as needed - but certainly to adjust Contrast, Brightness and Saturation - and Sharpness.

I haven't played with the video yet, but reemas' shots have certainly inspired me to give it a try. At VGA resolution, I'd intuitively expect the smart move would be to turn the Sharpness all the way down - meaning, no additional processing. The camera's wasting time doing that and then having to do frame compression with whatever's left over.

If you play with the two cameras in the Evo, you'll find that for nearfield - meaning, even beyond arm's length a little bit - the front camera is MUCH sharper. I think that's a natural side effect of the much smaller aperture of the front lens, but I was very surprised because while that rule of thumb holds true for film cameras, with digital cameras sharpness and detail are very much influenced by the light going to each pixel - while the back camera has over 6 times the points to distribute light to, it does seem to have a disproportionately larger aperture with which to accomplish that.

In color work, whether still photo or video, believe it or not, seeing detail comes from contrast first, resolution last - color/chroma and saturation even come before resolution. Lots of newcomers to HDTV find this counterintuitive, but it's quite true. Pop up the photos in this link, set them all to the same size, do a side-by-side comparison and see for yourself:
Colorfulness - Wikipedia, the free encyclopedia

With the cameras appropriately adjusted (both Evo and iPhone), I wonder if the shoot-out results would be materially different?

I also wonder if the camera's processes are in any way prioritized in the Evo, or if it's having to fight with other running processes for CPU resources. Hopefully so, because that could portend a software fix.

I'm not holding my breath on HTC doing anything about this. They're marketing numbers - try telling them you have some negative metrics against an iPhone, and I'd be surprised if you got anyone that cared the way you may want.

Apple knew that everything about the iPhone would be under scrutiny and skepticism - remember the days before the unveiling? Remember what Steve-o said the target market was in the keynote address? I sure do. So the iPhone wasn't designed by walking down the parts aisle with a checklist against what the other phones in the line had - each component, hardware or software - were targeted to catch an untapped segment of the smartphone market.

When you add up the leading edge work Apple has always done with video-related work (the QuickTime engine, Firewire, ... a long list! ...) and the fact that their phone has gone through refinement to sell one model only - and compare that to any other typical phone manufacturer like HTC, the expectations become obvious.

Scaling and video compression algorithms can be committed to a single off-the-shelf chip. If that's how the camera is being handled in the HTC, then its video woes for scaling and compression will not be fixable by any sort of update. If it's being handled by software/firmware - those algorithms are proprietary (read: they were purchased) so HTC won't be replacing them for free. (Again, we go back to the Apple model of focused development: they've had superior scaling and compression algorithms within their purview for a great many years as anyone using a Mac mini for a home theater PC, outputting vid via DVI to 720p/60Hz, 1080p/24Hz or 1080p/60Hz can attest, as can the owners of any Apple TVs.)

Well Ser reemas, I hope you're satisfied. It wasn't as if I didn't have enough to do already with my phone, now I have to go play with the video!

PS - If anyone's skimming, and my post is TLDR, maybe you'll notice this PS. It says to go back and read SporkLover's last post (#50) on the previous page.

 

[wrchad23]

I'll never understand why when someone brings up a shortcoming of someones phone they act as if they called their mother a fat, whore whale or something. I think this is a great test.

 

[mrspeedmaster]

hahhahahah, Cmon No iphone would win in Phandroid, us Android fans are more loyal than apple fans!! so w.e you say you are wrong!!
the Evo Rules, Android Rules, And apple and Iphone, could go to hahahhaha

How old are you? Just asking?

 

[misterzeno]

wow many epic posts in this thread, i will defiantly check with some of you when it gets time for me to get a DSLR

 

[reemas]

great responses! i will get outside to shoot some day/outdoor video today and will re do the test.

as for 720P, i can do that, but it will be pointless as this camera shoots BETTER in VGA (640x480).

i will flip a coin to randomize which is on the left and right again. but fwiw, the EVO compresses the crap out of the videos, so it might still be obvious which one is the EVO and which is the iPhone. we'll find out in a bit!

 

[EarlyMon]

great responses! i will get outside to shoot some day/outdoor video today and will re do the test.

as for 720P, i can do that, but it will be pointless as this camera shoots BETTER in VGA (640x480).

i will flip a coin to randomize which is on the left and right again. but fwiw, the EVO compresses the crap out of the videos, so it might still be obvious which one is the EVO and which is the iPhone. we'll find out in a bit!

If you feel it's fair - and I think it is - try setting ISO to non-auto, turn down sharpness, maybe bump saturation up. I just tried that ISO 200, n/c to brightness and contrast, and it didn't look too bad at capturing the desert mountains on a bright day.

And if you could - kindly kill off the other non-essential apps.

In any case - kindly note your settings for both cameras.

Thanks in advance for the retests!!!

 

[reemas]

If you feel it's fair - and I think it is - try setting ISO to non-auto, turn down sharpness, maybe bump saturation up. I just tried that ISO 200, n/c to brightness and contrast, and it didn't look too bad at capturing the desert mountains on a bright day.

And if you could - kindly kill off the other non-essential apps.

In any case - kindly note your settings for both cameras.

Thanks in advance for the retests!!!

ok i'll try to get all this right, but i don't think i can set the ISO for video.

also, FYI, i can choose H263 and MPEG4, but H263 is limited to 320x200 or something small and useless like that.

 

[EarlyMon]

ok i'll try to get all this right, but i don't think i can set the ISO for video.

also, FYI, i can choose H263 and MPEG4, but H263 is limited to 320x200 or something small and useless like that.

Doh! I bumped the camera selection when jumping around menus! Doh! B^/

 

[reemas]

ok i didn't have time to do everything, so i'll try again later today. this is a video shot outside in bright sunlight. i actually set the EVO to 720P.

couple notes: my right hand was shaking a bit in the beginning, so i actually shook my left hand a little more. as a result, you'll notice the iPhone 3Gs handles motion a LOT better than the EVO.

also the file is huge, about 90 MB. and the resolution is EVEN bigger, so you can see the two side by side. i'll get a better test out later today hopefully with both camera on 640x480 and outdoors. but don't hold your breath, because you know what it will look like.

www.svohra.com/web/test/test2.mov

 

[EarlyMon]

OK - just a minute.

I just recorded a video on my Evo, brought the 3gp file over via USB, converted the 3gp file to mp4 (essentially just changing the container, not the codec) and according QT, my frame rate is 28.63 fps.

And it looks way good.

Something's very wrong with that low fps result, if I read that correctly earlier.

PS - OK, posts said that that was for low light. Not sure how low. I'm shooting indoors, daytime, getting a steady 28 fps or better - at VGA.

Just shot the same thing in 720p - got an abysmal 9.77 fps.

Huh.

PS - I use MPEG Streamclip, have for some time, swear by it, so for those of you who need that sort of thing: http://www.squared5.com/

 

[EarlyMon]

Doh. This is a no brainer. I just noticed from your latest test that the iPhone does everything in H.264 (AVC, MPEG4 Part 10) vs. the Evo's MPEG4 Part 2 (ASP, note compatibility to H.263).

Nothing more to decide or test. H.264 is simply a superior codec.

In other news - while looking for H.264 support for Android, I did stumble across this:

CorePlayer on Android: playing H.264 Video | Droid Sector

So, I looked them up... Home-CorePlayer

Those following along will also want to see this:

CoreCodec Community Forum • View topic - CorePlayer for Android...

I read through that entire forum - very interesting.

Is it possible we'll have that Froyo? BetaBoy's comments did sound better as things progressed.

And just because I'm that way, yeah, I looked, and we're using the PacketVideo core in Android - What is Android? | Android Developers

So.

MAYBE we will be able to buy an H.264 _encoder_ for the phone, in future.

And then we might get closer to the iPhone's video performance.

Maybe. Might. Future.

BTW, I'll just say it again - Ser reemas, I hope you're thoroughly satisfied with yourself!!!

 

[reemas]

earlymon, thanks for all the info. why would i be satisfied ha ha?