GX200 vs. GX100 vs. R7 - Noise Reduction differences

[odklizec]

Some time ago I pointed on excellent chroma noise removal algorithm in R7/R8 cameras. Today I did a detailed GX100/GX200 ISO shootout and my preliminary comments and worries about integrating 12MP sensor and change in NR strategy in GX200 seems to be confirmed. The GX200 is noisier and the NR is much more aggressive than in its predecessor.

It will take me some time to prepare all relevant crops, but I would like to show you at least small example describing how superior is R7 type of noise reduction. Unfortunately, I currently don't have my R8 with me so I had to do this test with R7. However, the noise reduction in R7 and R8 is more or less the same. So here are two quick examples.

r7_gx100_gx200.jpg (510.87 KiB) Viewed 9344 times

r7_gx100_gx200_2.jpg (540.22 KiB) Viewed 7185 times

The GX100 and GX200 samples were only scaled down to R7 native size. All camera settings are the same, however, it's impossible to set some settings in R7. While in GX100/200 is used Sharpness -2, in R7 is used "Soft" settings. The white balance seems to be slightly different too (although all cameras were set to Cloudy WB).

It seems that R7 noise reduction outperforms both GX100 and GX200 noise reduction. R7 effectively reduces color noise in shadow parts and leaves only light grain with details. GX100 does only small amount of NR leaving some of the color noise but still preserving details. There is not dramatic difference between GX100 JPEG and DNG files (at least not at ISO400). However, as you can see in GX200 crops, this camera produces very crude NR artifacts and killing all tiny details.

In my humble opinion, Ricoh should consider integrating R7/R8 type of NR also in their other products, like GX or GRD. Maybe not as a replacement of current NR algorithm, but rather as a selectable option, as suggested in GRDII FW wishes. As a RAW shooter, I personally don't care much about out of camera JPEG. But I'm quite sure that a lot of people would like to see rather a bit grainy but much more detailed and free of color noise JPEGs. And this Rx type of NR is definitely the way to go!

[Grus74]

Thanks for taking the time to do this comparison Pavel! Just looking quickly at the pictures I'd say that my favourite is the GX100 in that it has the (in my opinion) best balance between noise and detail.

[odklizec]

You are welcome Grus! It's just a start There is more comparisons on the way.

You are of course right. The GX100 is well balanced mix of noise and details, which is good for later processing. There is slightly less details on R7 crops, but I think it's mainly because of less powerful lens. In my opinion, the R7 type of NR is something I would like to get if I don't need or want to mess with post processing. Just out of camera images good for a bit of cropping, resizing and resharpening. But what's really great about R7 NR are shadow and dark parts of high ISO photos. I simply love the left grain and details and no visible color noise in these parts, which are usually full of ugly color noise or terrible noise reduction artifacts. And this is, in my opinion, where the R7 NR really shines. It's just like NoiseNinja color noise removal, which is really great. OK not exactly as good, but really good anyway

[Prognathous]

Could it be the the lighting was different in the GX100 shot? The shadow behind the guitar seems to be missing.

Also, the GX200 is surprisingly bad. Is this the typical difference between it and the GX100? Does it fair better in RAW mode? As displayed here, this is really bad performance for ISO 400 JPEG (not that I care so much about JPEG, but it's interesting nonetheless).

Prog.

[odklizec]

Hi Prog,

The lightning was exactly the same. It's just the GX100 sensor I'm preparing complete GX200/GX100 ISO test but it will take me some time to prepare all crops. So this is just an example how good or bad is the ISO 400 performance in different Ricoh cameras. I agree that GX200 hi-ISO perfomance in JPEG is not the best in the class. But you can get much better results from RAW.

[Prognathous]

Yes, I know that RAW + Noiseware can deliver far better results. See my experiment here:

http://imapnet.com/boren/GX200_HighISO/

By the way, do you happen to have a color patches diagram (e.g. Kodak Q13 or Macbeth Color Checker)? It can be useful to include such a diagram in your test. Check out dpreview.com tests of Nikon P5100 and Sigma DP1 to see how a camera can deliver relatively low noise at high-ISO while still fail miserably when it comes to color rendition. This is an important enough aspect in my opinion that I actually prefer the high-ISO results from the 1/2.x sensor Panasonic TZ5 over both cameras.

Prog.

[mark1958]

I would be very interested to see a GX100 vs GX200 comparison of noise at ISO 400 and 800 (forget 1600 ---hopeless) made from a RAW capture. Based on your experience is the Noise the same or worse in the GX200.

[mark1958]

Looking at your guitar examples. The GX100 and 200 have much more detail than the R7 but the artifacts in the GX200 compared to the GX100 is much more apparent due to the aggressive noise reduction algorithm. This is really a shame. THe exposures are slightly different though.

[odklizec]

Hi Mark and welcome here! I already completed the ISO test, comparing GX100 and GX200 JPEG/RAW at all sensitivities and also with various JEPG settings. But I will have to find some time to make crops, merge them together and upload all relevant files. I hope to finish it all during this weekend.

From what I've seen, the GX200 noise level is definitely higher. It's not a problem in RAW at ISO 64-400. But it starts to be a problem from ISO800 and above. These high ISO shots seems to be much harder to clean from color noise than GX100 or GRDII files. Unfortunately, this is the price we pay for 12MP sensor.

[mark1958]

what do you use for your noise reduction for the GX files? I have used noise ninja for DLUX 2 files. I would really like to see some comparisons between the Panasonic LX 3 and GX-200. For me RAW is all i am concerned with.