ATI Driver for Touch Pro

[dr g]

Originally Posted by cornelious2

That being said there is processing between the sensor and the buffer and since the sensor cannot actually hold data it has to stay open the whole time while that processing happens. on highend cameras there is another dedicated processor that does this. on low end the main processor handles this, but either way there is video processing happening.

Not true, sensors can indeed hold data; charge is shifted out of the light-sensing pixels in various ways after exposure into transport and/or storage circuitry on the sensor. CMOS sensors have per-pixel readout capabilities, and are able to transport accumulated charge from the light-sensitive transistor to a buffer transistor to make electronic shutters possible (which is what makes our cell phone cameras possible).

Read about electronic shutters here:
http://www.caspegroup.com/How%20an%2...S%20camera.pdf

Now here's the thing ... the electronic shutter is likely a rolling-shutter type which can be prone to artifacting with moving objects based on the maximum readout speed of the sensor. This is, however, not the source of the blur we see. First, the blur we see is not the kind of artifacting caused by electronic rolling shutters -- those exhibit skewing and image distortions, not long-exposure type blur. Second, we are able in bright light to eliminate blur, which indicates that the maximum shutter speed of the camera is not the issue ... theoretically we should be able to use that same shutter speed in low light. It's just a question of what shutter speed the camera picks to use in various lighting -- this is a function of the camera software, the drivers in question -- no matter what they do -- would not affect this.

[dr g]

Originally Posted by hoddy4

yea, the only issues i have with the pro are its graphics performance (particularly slingbox) and the heating issue. i'm hoping that the graphics can be fixed by a driver (or maybe an optimization of slingplayer), but the heat issues may require a hardware redesign unless someone can come up with a clever rear cover that removes heat more effectively. from my perspective the pro and the ppc6700 have been the best of the htcs.

Actually I think a more efficient radio driver could solve the heat issue. If power consumption could be lowered significantly and radio heat better controlled, we might be able to avoid the cutoff.

[jakecramer]

this driver made a HUGE difference in my camera. it made it way less laggy and it takes much better pics than before. if i was shaking slighty before the picture was blurry but now its perfectly clear and also it takes the picture much quicker than before.

[Mutiny32]

Originally Posted by dr g

Unfortunately everything in the paper is irrelevant to the discussion because it happens AFTER the exposure is done and the data is read out and placed into the pipeline. The data pipeline should NEVER affect exposure time in a properly functioning camera. Precise control of exposure time is critical to photography in general.

The D3D driver on the Qualcomm MSM chipset contols camera function. That's it. That's all there is to it. Your argument is null. Something in these drivers fixed camera exposure metering and thus caused the problem to be fixed.

That something is probably the capability to accurately measure exposure in a timely fashion and adjust ISO and exposure properly.

[WarChild]

Originally Posted by dr g

Not true, sensors can indeed hold data; charge is shifted out of the light-sensing pixels in various ways after exposure into transport and/or storage circuitry on the sensor. CMOS sensors have per-pixel readout capabilities, and are able to transport accumulated charge from the light-sensitive transistor to a buffer transistor to make electronic shutters possible (which is what makes our cell phone cameras possible).

Read about electronic shutters here:
http://www.caspegroup.com/How%20an%2...S%20camera.pdf

Now here's the thing ... the electronic shutter is likely a rolling-shutter type which can be prone to artifacting with moving objects based on the maximum readout speed of the sensor. This is, however, not the source of the blur we see. First, the blur we see is not the kind of artifacting caused by electronic rolling shutters -- those exhibit skewing and image distortions, not long-exposure type blur. Second, we are able in bright light to eliminate blur, which indicates that the maximum shutter speed of the camera is not the issue ... theoretically we should be able to use that same shutter speed in low light. It's just a question of what shutter speed the camera picks to use in various lighting -- this is a function of the camera software, the drivers in question -- no matter what they do -- would not affect this.

Wow +1 for the homework

[dr g]

Originally Posted by Mutiny32

The D3D driver on the Qualcomm MSM chipset contols camera function. That's it. That's all there is to it. Your argument is null. Something in these drivers fixed camera exposure metering and thus caused the problem to be fixed.

That something is probably the capability to accurately measure exposure in a timely fashion and adjust ISO and exposure properly.

BS. The problem is not fixed by these drivers. There would be dancing in the street if it did. What is probably happening is that on some underperforming devices, this driver accelerates the display of the live camera preview. This solves a bottleneck between the processor and the display, making preview and focusing more responsive, thus improving usability and user success rate. But the real issues are not solved, as responsiveness is still slow overall and shutter speeds are too slow for action/candid shooting, especially in low light.

I'm willing to bet that no one who reports their camera is "fixed" by this driver can do the kind of candid shooting we are expecting and lamenting the device's incapability of.

[brazen]

Using Sprint TP stock rom, just received in mail 2 days ago, and it's already set to use 'd3dm_ati.dll'

[Raging Idiot]

Originally Posted by Mutiny32

The D3D driver on the Qualcomm MSM chipset contols camera function. That's it. That's all there is to it. Your argument is null. Something in these drivers fixed camera exposure metering and thus caused the problem to be fixed.

That something is probably the capability to accurately measure exposure in a timely fashion and adjust ISO and exposure properly.

A few months ago when I got this phone I did some extensive testing with the camera software along with the various tweaks that existed at that time. Nothing that existed worked.
I remember mentioning here that I found the best option for faster shooting was simply changing the "Metering Mode" option (located in the standard camera menu) to be "Average" rather than the default of "center". I don't have time to read this entire thread and it is quite likely this has already been covered. Just figured it possible that someone made an adjustment to the "metering mode" somewhere in one of the fix-it files.

When it comes to camera speed everything can effect everything so don't rule anything out just from what you read on the internet. Definitions might be fun to quote, but it's theory... not law. Much like there can be several reasons your storage card r/w speed can change, so too the camera. Any nay sayers will be promptly hung in a noose of logic for the world to see...so don't.

In the end, all that matters is results. I tend to favor the skeptics, as they tend to be right, although sometimes outnumbered.
Reminds me much of when the 6800 first came out and I had argued that you couldn't upscale the processor using anything that worked with the 6700 because of the extremely different processor. However, dozens of people, (mostly rather young/naive) were saying they were effectively upscaling the dual-core processor using the incompatible software. LOL
So beware the power of suggestion and any placebo effect. It's frustrating when you have a few people say something works and then you scientifically test it and find out it was a waste of time.
It is most likely (if any significant improvements are possible) that the improvements will come from porting over the software from other similar devices. Until then, I'll deal with just having the Metering Mode set to Average.

[lgmayka]

So is there a verdict yet as to whether to use the ATI or the Omnia driver?

[dr g]

Originally Posted by lgmayka

So is there a verdict yet as to whether to use the ATI or the Omnia driver?

For me the verdict is ATI. The Omnia driver didn't work too well for me.