[Ray Pulley]

K7¸ß¸ÐÌá¸ßÁËÂð£¿K7ÓëK20D¸ß¸Ð¿ØÔë±íÏÖÔ*ͼPK£¬Ã¨É÷Èë - ÎÞ¼ÉÂÛ̳

Pre-release firmware, so this is not really fair, but it reminded me how good the K20D is at 1600.

Apologies if this is a dupe.

Ray

[GLXLR]

Looks fake to me. Why would the sensor perform worse than the predecessor? (yet at same resolution)

[soalle]

Pre-production firmware!

[Ray Pulley]

Looks fake to me. Why would the sensor perform worse than the predecessor? (yet at same resolution)

Because it is not the same sensor.

It is a 4 channel readout, which is needed to get the frame rate up to the 5fps. This has noise implications that a 2 channel readout does not have (the K20D is 2 channel).

We also have all new circuits and a new Prime processor, new electronics insise (different noise and inerference charachteristics), different wiring, a new SR mechanism, new algorithms, etc..

This K-7 also does not have the final firmware and processing algorithms, and also may not even have final production circuit boards in it.

There is no surprise about it being a bit noisier than the K20D.

The initial K20D images were noisier than the final production release as well. In fact, IIRC, quite similar to what you see here on the K-7.

Ray

[nosnoop]

Pre-release firmware, so this is not really fair

Not only is it not fair, it is quite pointless, really.
Early firmware output has very little bearing as to what the final firmware output would look like. Remember the horribly noisy first photos leaked for the K10D?

[Ray Pulley]

Not only is it not fair, it is quite pointless, really.
Early firmware output has very little bearing as to what the final firmware output would look like. Remember the horribly noisy first photos leaked for the K10D?

Seems as if we have lots of "forum police" here nowadays. It is starting to feel a lot like the "other" forum here these days.

Have you been objecting to the thousands of pointless threads speculating about the features and performance of the K-7, many even before ANYTHING was known for sure?

If you think a thread is pointless, feel free to NOT read it.

Ray

[joerg]

I know that this is not final firmware, but wow, the k20d did quite a bit better here. Pentax has fallen behind just a bit in noise levels so for this camera to go backwards or even stay the same might not be a good thing. Also seems to me that a firmware update will have less to do with sensor performance and more to do with changes to noise reduction.

[Ted1212]

The K7 noise seems less destructive, look at the blacks under the flash for example. The K7 is usable up to 2500 on these.

[ricardobeat]

Because it is not the same sensor.

It is a 4 channel readout, which is needed to get the frame rate up to the 5fps. This has noise implications that a 2 channel readout does not have (the K20D is 2 channel).

We also have all new circuits and a new Prime processor, new electronics insise (different noise and inerference charachteristics), different wiring, a new SR mechanism, new algorithms, etc..

This K-7 also does not have the final firmware and processing algorithms, and also may not even have final production circuit boards in it.

There is no surprise about it being a bit noisier than the K20D.

The initial K20D images were noisier than the final production release as well. In fact, IIRC, quite similar to what you see here on the K-7.

Ray

Those are visibly not production-quality shots. You wouldn't accept that ISO 6400 noise from a P&S. Wait until someone has a production model in his hands, with final firware, to make your comparisons.

According to most reviews the 4 channel readout should offer better noise performance. People "in the know" have mentioned that it can even make images too buttery for a pentaxian

[er1kksen]

Seems as if we have lots of "forum police" here nowadays. It is starting to feel a lot like the "other" forum here these days.

Have you been objecting to the thousands of pointless threads speculating about the features and performance of the K-7, many even before ANYTHING was known for sure?

If you think a thread is pointless, feel free to NOT read it.

Ray

I think he was referring to the comparison between images from a K20D and images from a K-7 with preproduction firmware as pointless, not this thread. In fact, he might agree that this thread has a useful point indeed: to point out that the mentioned comparison is in fact pointless, hopefully staving off any underinformed naysayers who might jump on the comparison and start bashing the K-7 as a result.

[KungPOW]

Some speculation follows...

Pentax may have the K-7 sensor and software "tweaked" in the beta test bodies to do next to no noise reduction. They might even have that part of the software turned off.

Why?

So they have a bunch of testers send back all their high ISO shots giving Pentax a large sample of typical High ISO shots to use to adjust the noise reduction software.

I expect software/image engineers spend hours with thousands of images like the ISO 1600 K-7 image here and they adjust and tweak until they get the best overall results with a wide variety of images from many different shooters with many different styles. I'm not sugesting they tweak each individual image, but rather run them through the software in bulk, and review the results.

Then when everone is happy with how the software cleans the image, it becomes firmware.

Again, I don't know anything about the process, this is just speculation.

[er1kksen]

I think it's pretty obvious from the comparison shots that that preproduction K-7 is doing a LOT of noise reduction, actually. And not the type I'm looking for. Fortunately, by the time I'm able to afford a K-7 I should also have a more capable computer and be back to shooting RAW.

[robgo2]

Seems as if we have lots of "forum police" here nowadays. It is starting to feel a lot like the "other" forum here these days.

Have you been objecting to the thousands of pointless threads speculating about the features and performance of the K-7, many even before ANYTHING was known for sure?

If you think a thread is pointless, feel free to NOT read it.

Ray

So now you're policing the police. This thread is a perfect example of how nasty rumors get started. Not one of us has any idea how these images were made and whether they are genuine. You have every right to post them, but others have the right to call you on it.

Rob

[ricardobeat]

I think it's pretty obvious from the comparison shots that that preproduction K-7 is doing a LOT of noise reduction, actually. And not the type I'm looking for. Fortunately, by the time I'm able to afford a K-7 I should also have a more capable computer and be back to shooting RAW.

It looks compressed actually, the blur makes it look less noisy, but you can notice there is absolutely no chroma noise reduction going on.

[jaad75]

I think it looks better than K20D in these samples, the problem is that you people don't know where to look... K20D cuts deep shadows very quickly and there is absolutely no detail to recover, K-7 seems to have more DR over there. Take a look at blue and red channels in both cameras - K-7 has much less blue noise, so it's gonna look much better with boosted blue channel when compensating for tungsten (which is where you would probably use higher ISO anyway). I can already make this early pre-production sample look better than the one made with K20D - I'm really thrilled to see what they can archive when they finally tweak the software and hardware to the optimal level...