GH2 DxOMark is worse that GH1!

N

noirist

Well-known member
DxO Labs tested the GH2 sensor and it's worse than the GH1 sensor in every measurement! The GH1 is 64 overall, 21.6 bits color depth, 11.6 Evs dynamic range, and 772 ISO. The GH2 is 60 overall, 21.2 bits color depth, 11.3 Evs dynamic range, and 655 ISO. Maybe it's because they made the pixels smaller so more light is lost in between the pixels? What is Panasonic thinking????

http://www.dxomark.com/index.php/en/Camera-Sensor/Compare-sensors/%28appareil1%29/630|0/%28appareil2%29/677|0/%28onglet%29/0/%28brand%29/Panasonic/%28brand2%29/Panasonic
 
I don't think its a joke. When you try to squeeze more pixels onto a sensor, you must make the pixels smaller, and the sensor doesn't absorb as much light. Either the individual pixels can't absorb as much light relative to their surface area, or there's more space between the pixels that doesn't absorb any light, or some combination of the two. Take a look at the Nikon D3s, which has the most amazing low light capability because it has relatively few very large pixels on a giant full frame sensor. Panasonic made the GH2 pixels smaller so they could put more of them on the sensor for marketing purposes and now they are burned. It looks like the high ISO claimed for the GH2 is just a software trick.

P.S. The GH1 pixel pitch is 4.31 micrometers and tested ISO 772. The GH2 is 3.63 and tested ISO 655. The D3s is 8.4 and tested ISO 3253 (!).
 
Last edited:
DxO doesn't measure FPN, just noise. If it's got rid of the FPN, then I'd be happy with the other improvements.
 
That's not uncommon when they increase pixel count, decrease size of pixels. Canon 40D scores higher than the 50D. It is the trade off for higher resolution. GH2 also has much faster, more powerful processing so it will be interesting to compare overall quality.
 
Wow DxO Labs just tested the Pentax K-5 as well: 83 overall, 23.7 bits color depth, 14.1 Evs dynamic range, 1162 low-light ISO. For not much more than the GH2. I wonder if the 1080p25 from the K-5 is any good?
 
Calm down, look at the graphs, nothing has changed they are the same pretty much, not sure why really that they have lost points its so small the difference. Also, note, the Canon60d, 7D are only at 66, the Panasonic is close, but really as we all know who have GH1's noise is much better on the GH2 and for anything but the largest prints in stills from raw you will not notice any difference.:Drogar-Happy(DBG):
 
We'll see.
We have new firmware coming soon for K5.
And as I have all tools for Pentax, most probably it could be hacked.
Right now Pentax do not have full manual control of shutter speed in aperture priority mode for video.
 
noirist said:
Wow DxO Labs just tested the Pentax K-5 as well: 83 overall, 23.7 bits color depth, 14.1 Evs dynamic range, 1162 low-light ISO. For not much more than the GH2. I wonder if the 1080p25 from the K-5 is any good?
Click to expand...

Loved Pentax for years but had to move on once video came to dslrs.
 
Last edited:
so what if we gave up a little noise performance for some more pixels? if the *color banding*(it's not FPN people..) is gone, and the increased number of pixels give you a sharper picture(smaller steps on the aliased edges) then who cares about a little noise?

Have you seen how noisy film is? It's PART of the look that gives film the *film look*.

How about we stop reading all these sites and wait to see how the video actually looks? I bet it's just fine.

the only thing I'm worried about is codec breakup under detail/movement.
 
Vitaliy Kiselev said:
We'll see.
We have new firmware coming soon for K5.
And as I have all tools for Pentax, most probably it could be hacked.
Click to expand...

Great to hear.... the more hackable cameras we have, the better.... and if that DR range is usable on the video, it could be very good. Almost film like.
 
The K5 is seeming like a really good choice. I'll take 25p, MJPEG I just like better, the manual control thing I heard about, Back when i had a d90, I hated that there was no iso/shutter control, but even that there wasn't a readout. I heard that the k5 can display a readout during video mode of the iso and shutter. I could try this thing out. If it is hackable, I'll take 25p mjpeg in a flash.
 
noirist said:
Panasonic made the GH2 pixels smaller so they could put more of them on the sensor for marketing purposes and now they are burned.
Click to expand...

Maybe they put more megapixels on the GH2 because they think the new Extra teleconversion feature is going to be a big hit. The new GF2 has the feature. However, the GF2 can only use it in 720p mode because the sensor does not have as many pixels as the GH2. Just a thought.

They might have actually said, "we can trade some image quality for features and get away with it because the software processing is slightly better as well". I think they are neglecting the fact that most really picky people shoot RAW. The software improvements are not going to help with RAW processing.
 
Fpn

Fpn

Svart said:
if the *color banding*(it's not FPN people..)
Click to expand...


Yes it is.

Unless you're talking about bit-depth banding, which is a codec issue, possibly a dithering one.
But when most people talk about the "banding problem" they're referring to FPN, which exhibits itself worst as straight lines of strong chroma (noise) color, more visible in flat low-light areas.

And it won't be "gone" even if it is somehow improved (doubtful on a sensor / RAW level, probable on a processing level): FPN is a fact of CMOS sensors. They all have it to some extent, period.
 
Last edited:
dasbin said:
Yes it is.

Unless you're talking about bit-depth banding, which is a codec issue, possibly a dithering one.
But when most people talk about the "banding problem" they're referring to FPN, which exhibits itself worst as straight lines of strong chroma (noise) color, more visible in flat low-light areas.

And it won't be "gone" even if it is somehow improved (doubtful on a sensor / RAW level, probable on a processing level): FPN is a fact of CMOS sensors. They all have it to some extent, period.
Click to expand...


Nope. You are also mistakenly mixing those up too. Most folks around here are speaking of the colored banding noise, the yellow/purple bands that tend to show up in dark areas. that's color banding noise, not FPN. FPN shows up as thin black lines in checkerboard pattern. they arrange in similar patterns to each other but most folks see the color banding, not FPN.

http://en.wikipedia.org/wiki/Screen_door_effect

And you'll find "banding noise" here:

http://www.cambridgeincolour.com/tutorials/image-noise.htm
 
This won't stop me from getting GH2, but I demand cheaper price. the 14-140 kit at $1200!!!
 
Last edited:
You must log in or register to reply here.
Back
Top