Hey theSuede, I think you may have posted the same shot twice above. The cloud position matches in both shots where in the original comparison they are different. If you can update the post it would be easier to compare..they are a lot closer.
tsdevine wrote:
Hey theSuede, I think you may have posted the same shot twice above. The cloud position matches in both shots where in the original comparison they are different. If you can update the post it would be easier to compare..they are a lot closer.
-Tim
Thank you very much, my mistake when copy-pasting the links... images are correct now, one of each, not two of the same. Sometimes brain-finger coordination breaks down :/
(BTW, the top/first image is the Canon image...)
No worries, you don't have the market cornered....I know I've made mistakes posting multiple links. I'm so focused on what I'm writing that I don't notice whether all the links were actually right.
Great stream shot Jim, picked up the DP2M and later the DP1M....but it was after all the leaves dropped this fall. I love this type of photography and I keep waiting for things to green up around here so I can shoot it with more foliage. Your shot is great and makes me look forward to it even more!
theSuede wrote:
Thank you very much, my mistake when copy-pasting the links... images are correct now, one of each, not two of the same. Sometimes brain-finger coordination breaks down :/
(BTW, the top/first image is the Canon image...)
/hangs head in shame/
That's funny, I stayed away from your original post-processed comparisons cause it felt like a trap - they looked exactly the same! "How did he do that?", I thought. I was amazed at your cloning and sharpening skills to make them look identical. Now that you reposted them, I will say that the Sigma file does still appear more natural to me, the enhanced detail on the sharpened Canon file doesn't look quite as clean. But your point is valid nonetheless.
PP options on 5D3 files are much easier to execute, and it's infuriating that Sigma forces us to use their own SPP software to convert RAW images. I haven't taken the time to figure out how much SPP can really do.
tsdevine wrote:
Great stream shot Jim, picked up the DP2M and later the DP1M....but it was after all the leaves dropped this fall. I love this type of photography and I keep waiting for things to green up around here so I can shoot it with more foliage. Your shot is great and makes me look forward to it even more!
-Tim
thanks Tim. I hope you get out there with them soon. How are you liking the DP1M?
I haven't shot the DP1M as much yet, but it doesn't have the edge to edge sharpness of the DP2M...at least mine doesn't. The center is tack sharp, as is maybe 2/3rds of the frame. But as you get to the edge and corners, it loses that bite. It also seems much more prone to CA.
Back in the film days I had 2 Ricoh GR1's, I see Ricoh just came out with the latest GR digital. That lens looks sharper on the edge and corners...wonder how that lens would have performed on the DP1M. From what I understand though, the Merrill sensor is more sensitive to the incident angle...so who knows, that may have made the lens design more of a challenge. May also explain the sharpness advantage on the edge and corners of the DP2M and DP3M.
I'm more of a wide angle shooter....but I started with the DP2M because of the samples I was seeing. Decided to pickup the DP1M anyway.
I will play with my 6D and 40STM (the only prime I have close in FL) with my DP2M this weekend and compare. Good excuse as any to head out somewhere and shoot.
there's no question that the resolving power of the Foveon sensor can compete with much higher-megapixel Bayer sensors...but IMO it just falls apart on everything else: noise, shadow posterization at base ISO, blotchy skin tones in difficult lighting and in shadows, really horrendous workflow. I've brought this up in the past, and to be honest I started to question my workflow skills at the time, but it turns out that I wasn't alone with my assessment.
with all that said, I really hope that Sigma eventually perfects the technology. I wonder, though, is that why should the primary colors in separate layers? why couldn't Sigma distribute the primary colors in all layers so Red channel would not have to suffer from too much noise?
p.3 #10 · Sigma DP2M v Canon 5D3 - resolving power
As the owner of a 5D3, an X100 and a DP2M, I have to say that I find myself leaving the 5D3/21 distagon combo at home in favor of the DP2M and X100. If I want wide I stitch. But as far as the OP's title regarding resolving power is concerned, I find that with care, the files are quite similar. However, the unique 'foveon look' ( yes, even with the sharpening dialled back) still gives the appearance of greater clarity to my eyes. And its not just me, most of my mates ( including a D800 owner) are pretty much in awe of the DP2Ms output, particularly for detailed landscapes.
p.3 #11 · Sigma DP2M v Canon 5D3 - resolving power
dalegaspi wrote:
with all that said, I really hope that Sigma eventually perfects the technology. I wonder, though, is that why should the primary colors in separate layers? why couldn't Sigma distribute the primary colors in all layers so Red channel would not have to suffer from too much noise?
The colors are detected based on photon diffusion depth, so you can't detect different colors at the same depth. Distributing the primary colors in the same depth means you need to put them next to each other, which gives you the same thing as an ordinary Bayer array sensor.
The concept behind the Foveon technology is to detect all 3 primaries at the same location (in each pixel), which gives better chroma resolution, but less accurate colors (and more noise overall).
p.3 #12 · Sigma DP2M v Canon 5D3 - resolving power
AhamB wrote:
The colors are detected based on photon diffusion depth, so you can't detect different colors at the same depth. Distributing the primary colors in the same depth means you need to put them next to each other, which gives you the same thing as an ordinary Bayer array sensor.
The concept behind the Foveon technology is to detect all 3 primaries at the same location, which gives better chroma resolution, but less accurate colors (and more noise overall).
I wonder if this is something more R & D can improve upon or is it the best it can be right now.
p.3 #13 · Sigma DP2M v Canon 5D3 - resolving power
As theSuede and Samuli point out there is a lot more to be had in sharpening Bayer images if shot well, if pixel level detail is the feature that you prioritise. This will most likely become a bigger issue for Adobe soon, they are way behind on it..they provide the tools but everyone needs to evolve their own methods. There is nothing wrong with 100% appreciation either.
And doing no sharpening at all on the Canon image means nothing to the final image appearance.
The Sigma waterfalls look very surreal indeed. The 2nd image has lots of hot pixels in the high tone rocks, and water when it is near mist is always soft, here it looks like a series of small white balls.
And finally, the wide zoom is not one of Canon's finest moments. This is the kind of scene the 50MP excels at. I doubt it gets better in (DSLR) 35-50mm lenses regarding what you are wanting to show here - raw 40+ lpmm performance.
But really, any good prime or a 35-70 Leica or Contax will get you excellent results.
Thanks for putting these us, very interesting. The images look almost too sharp for printing.
p.3 #14 · Sigma DP2M v Canon 5D3 - resolving power
jfwoodman wrote: That's funny, I stayed away from your original post-processed comparisons cause it felt like a trap - they looked exactly the same! "How did he do that?", I thought. I was amazed at your cloning and sharpening skills to make them look identical. Now that you reposted them, I will say that the Sigma file does still appear more natural to me, the enhanced detail on the sharpened Canon file doesn't look quite as clean. But your point is valid nonetheless.
PP options on 5D3 files are much easier to execute, and it's infuriating that Sigma forces us to use their own SPP software to convert RAW images. I haven't taken the time to figure out how much SPP can really do. ...Show more →
Well, it unintentionally BECAME a trap, so I guess you did the right thing... But anyway.
My sharpening was executed on an already converted and compressed image, sharpening a 16-bit TIFF from a good raw converter is very different. In several parts of the Canon image, you see that the sharpening has amplified the jpg-artifacts way over the visibility threshold, and this adds to the "unnatural" look.
What most Bayer cameras are really missing is a really good raw-conversion, and GOOD sharpening. Adobe's solutions are sadly lacking in both - they're mainly geared towards cutting as many corners as possible and using as low precision as possible - to gain speed. Not that LR is a speed monster in any way, but with their coding quality doing it "correctly" would mean that the client probably dies from old age before you can deliver the images.
p.3 #15 · Sigma DP2M v Canon 5D3 - resolving power
Which is why I dont use Adobe apart from using PS for some cloning/generic pp stuff.. And LR as organizer.
And about Foveon possible to be improved. Im quite positive it can be. I strongly suspect Sigma doesnt have Sony class tech, which in result is generating more heat and noise than is needed. Tho not so sure if colors can be improved, but who knows, maybe they will find some way.
Easiest way would be probably trying to figure out something different with same results as Foveon X3. I think it can be done, but it depends if some manufacturer will be crazy enough to try it. But as Fuji showed, some are not so afraid of new.
p.3 #16 · Sigma DP2M v Canon 5D3 - resolving power
Thanks for the comparison. The images speak for themselves,and there has also been a lot of interesting comments.
However, for the purpose of comparing detail or resolution, the method of comparison is very important.
What do we actually see in the detail comparisons? From the relative sizes, it looks like the 5D3 image has been downsampled to match the DP2M pixels.
Downsampling just a little with a bicubic algorithm is about the worst thing you could to concerning perceived detail crispness.
Could you try upsampling the DP2M crops to match the 5D3, and then apply appropriate sharpening to the 5D3?
Besides, and I don't think anybody has mentioned it in this thread, the Sigma processing is cheating you by applying Local Contrast Enhancement with a large radius. Something like USM with 15 pixels radius and strength 20. In some scenes, this looks like "exceptional clarity", but in other scenes it looks like that ugly "HDR look" that was so popular around 2007. I can't believe that people really are falling for that trick. I mentioned it in the dedicated DP Merrill thread and tried to stay out of there since, but given this is a new thread with the purpose of comparison, I want to point it out again:
p.3 #17 · Sigma DP2M v Canon 5D3 - resolving power
Almost every time I see photos from Sigma's most recent sensor, they look great, and almost every time, a long discussion follows where a herd of photographers explain that in reality, they don't look as great as they appear, and if the files from camera C or N get enough attention in PS or another editing software, they will in fact look greater.
So probably there's something wrong with my eyes or my brain, which wouldn't surprise me after staring through a viewfinder for 40 years straight, but I still consider buying one or two of those Sigmas. During my wildest, sickest moments, I even consider an SD1 and a couple of primes.
Alternatively, I could go see a doctor and ask for some Canikon medicine
p.3 #19 · Sigma DP2M v Canon 5D3 - resolving power
alundeb wrote:
Besides, and I don't think anybody has mentioned it in this thread, the Sigma processing is cheating you by applying Local Contrast Enhancement with a large radius. Something like USM with 15 pixels radius and strength 20. In some scenes, this looks like "exceptional clarity", but in other scenes it looks like that ugly "HDR look" that was so popular around 2007. I can't believe that people really are falling for that trick. I mentioned it in the dedicated DP Merrill thread and tried to stay out of there since, but given this is a new thread with the purpose of comparison, I want to point it out again:
Like you say, the performance of the Merrill's is still great with regard to resolution but the huge influence of these processing tricks cannot be overlooked. At smaller sizes up to native size, the SPP processing gives a false sense of detail. Take a foveon image and interpolate it up to a very large size and it falls apart fairly quickly as compared to FF 20+MP bayer images.
p.3 #20 · Sigma DP2M v Canon 5D3 - resolving power
Damn those sneaky Sigma people! I for one have been fooled into thinking that this little camera can outgun my 5D3 and all because of the false appearance of resolution and sharpness created by those cunning SPP programmers. I shall sell my DP2M immediately!
That's funny, I stayed away from your original post-processed comparisons cause it felt like a trap - they looked exactly the same! "How did he do that?", I thought. I was amazed at your cloning and sharpening skills to make them look identical. 
How are you liking the DP1M?