Werner_Utsch wrote:
@alundeb
No, you can not get the the file quality of the DPM's by PP Bayer files!
It's not just enhanced local contrast etc..
Best you can do is "mime"the files quality, but you don't get real sharpness and detail by doing so.
The Fovenon does not work with software tricks.

Unfortunately that VERY untrue.

The Foveon principle involves some very heavy lifting in software, the red channel is actually more blurred than in a Bayer-base AA filtered sensor. Trying to sharpen this while still keeping per-pixel signal integrity is one of the hardest things to accomplish with Foveon files, and also the main quality difference between Sigmas own software solution and other raw converters.

So the "Foveon sharpness" is indeed very much a software solution. The only part where Foveon is noticeably better is resolution and pixel contrast in blue - but this does make a large difference in many situations. Unfortunately it comes at the cost of abysmal color accuracy and several times stronger noise errors.

One side is an untouched pixel crop from a Merrill Foveon sensor, the other side is a processed D800 crop.
Which is which and how do you see it?

Right is Sigma. The left has Bayer interpolation color artifacts visible in the smaller license plate numbers and the brick wall texture (flecks of color not present in the right). The right has more magenta/green blotching in the midtones.

Also: nice stereo pair

mpmendenhall wrote:
Right is Sigma. The left has Bayer interpolation color artifacts visible in the smaller license plate numbers and the brick wall texture (flecks of color not present in the right). The right has more magenta/green blotching in the midtones.

Also: nice stereo pair



Here is from the D800 processed with the "Merrillizer" recipe, and one with downsizing with bicubic to the Foveon pixel count and sharpening like is normally done in comparisons.

Now you can see how much of the difference you see around on the web is Bayer or Foveon, and how much is processing

How long before the Adobe dudes put something together for web sharpenening? Most of the plug-ins are poor also. People want to shave with image edges these days.

Am I alone in actually enjoying the workflow? For me, the modern photography experience is 50% about capturing the image, often involving travel to interesting places in the good company of like minded friends and 50% about getting back to base and getting stuck into the process of extracting interesting or pleasing images from the data I have collected. I may not be particularly good at either but I certainly enjoy both. Now with the DP2M there are a few more steps involved but each one of these steps is full of new and interesting possibilities for me to experiment with, thus enriching the experience. If I were a pro working to a deadline and had to produce files that were as near perfect as possible then I suppose I may be frustrated, but as a pure amateur who enjoys PP as much as capture its a non issue.

Chris: just wait until you have a hefty backlog...

I can enjoy the post processing too. I just don't have time to sit as much as I would like and need to in order to catch up. Some cameras and raw converter set ups require a much lesser time in post pro, like the Ricoh GXR M > RPP > PS > Print.

@theSuede

Of course, all raw data are software processed, the Foveon maybe heavy. But that's not the point. Detail is detail and no software in PP (and i was talking about PP i.e. Photoshop et al) can create detail that was not there beforehand. That's untrue?

Some Bayer cameras have a lot of detail. The Merrill's have a lot of detail, who is denying this and why? (is it just a software illusion?)

In this regard the comparison with a D800 is a good one and speaks for itself. You should, however, compare all aspects of the two cam's.

Chrissearle wrote:
Am I alone in actually enjoying the workflow? For me, the modern photography experience is 50% about capturing the image, often involving travel to interesting places in the good company of like minded friends and 50% about getting back to base and getting stuck into the process of extracting interesting or pleasing images from the data I have collected. I may not be particularly good at either but I certainly enjoy both. Now with the DP2M there are a few more steps involved but each one of these steps is full of new and interesting possibilities for me to experiment with, thus enriching the experience. If I were a pro working to a deadline and had to produce files that were as near perfect as possible then I suppose I may be frustrated, but as a pure amateur who enjoys PP as much as capture its a non issue.


Here´s my dilettante input.
50% is the capture with travels, people, places, smells, tastes, lights... Sometimes a controlled macro indoors.
50% is make the shot real or completely a new image.

I don´t consider myself a photographer but a maker of images because I do photography on a regular basis over a period of less than 10 years. But I do illustration and painting (and design) for almost 30 years.
I don´t want to make painting with photography; that´s ridiculous. But my mind is the mind of visual artist who uses tools. But, I do love photography-photography. The eye and the camera; simple.Along with the digital workflow I´m currently learning the wet collodion process, because I love the error, the chance of the analog processes. Doing by hand and produce unique objects is fascinating.
Again; not to mimic the XIX images (I will not dress like a cowboy ,-).
I work for/with the graphic industry for decades so I like transforming processes with machines.
Oh; and love optics and all other knowledge around cameras and photography; modern or earlier.

Why am I attracted to Sigmas?
Two reasons: I like to start with the best quality available to maintain or not the basis. Also, good RAW files are the "ink" that support extreme manipulations if needed.
The other reason is that I love the different. Foveon is the different. I admire people who "think different" and respect the risk.

Work, work, work.

Sorry if this out of context of this post but I found the process talk interesting.
Regards.

Werner_Utsch wrote:
Some Bayer cameras have a lot of detail. The Merrill's have a lot of detail, who is denying this and why? (is it just a software illusion?)


I hope I didn't come across as denying that the Merrills have a lot of detail. They do, and evidently the best for a compact camera and APS-C sensor.

What I want to debunk is:

1) "Foveon images have an exceptional 'Clarity' ". That is only a cheap trick with large radius LCE that doesn't even have anything to do with necessary sharpening to counteract diffusion.

2) "Even Bayer sensors with higher resolutions downsampled to 15 MP cannot match the pixel level sharpness of a 15 MP Foveon file". That is untrue. The key is how you resample and sharpen. Bicubic downsampling will preserve softness. What you can do instead is to use Nearest Neighbor downsampling. It requires that you first interpolate bicubic to a higher pixel count that is a factor of a whole number of the Foveon sensor's native pixel count. Otherwise you get irregular artifacts from NN.

Thanks Alundeb, theSuede, and Mike for "Merrillizing" some of the Foveon hype.
And before anybody flames me, I still think it is a nice camera.
Have some Foveon fun!



alundeb wrote:
Werner_Utsch wrote:
Some Bayer cameras have a lot of detail. The Merrill's have a lot of detail, who is denying this and why? (is it just a software illusion?)


I hope I didn't come across as denying that the Merrills have a lot of detail. They do, and evidently the best for a compact camera and APS-C sensor.

What I want to debunk is:

1) "Foveon images have an exceptional 'Clarity' ". That is only a cheap trick with large radius LCE that doesn't even have anything to do with necessary sharpening to counteract diffusion.

2) "Even Bayer sensors with higher resolutions downsampled to 15 MP cannot match the pixel level sharpness of a 15 MP Foveon file". That is untrue. The key is how you resample and sharpen. Bicubic downsampling will preserve softness. What you can do instead is to use Nearest Neighbor downsampling. It requires that you first interpolate bicubic to a higher pixel count that is a factor of a whole number of the Foveon sensor's native pixel count. Otherwise you get irregular artifacts from NN.

Considering that with Sigma you get way higher real resolution, you can print larger. So printing as if you had 24 mpix camera isnt problem, actually neither is limit 36 mpix..

Surely its enough for really detailed A3 (and A2 without trouble).

Resolution isnt problem for Sigma.

One from today:


_SDI0466 por John Loves, en Flickr

Yesterday sucked. I had a great two hours shooting in the early morning and spent the rest of the day in front of the computer sorting and processing the files. NOT creative processing......selecting the keepers to bring into LR4, doing basic expossure and color adjustment, saving as TIFFS, deleating the rest from my harddrive and importing the TIFFS into LR4. Now, the next day I can finally get around to really seeing what I have and working on them.This,is the fun part after the joy of shooting them. The rest is a 100% pain in the ass, no fun......I just don't have the time to dick around with these files. If Sigma doesn't become proactive in bringing about some usable software for photographers who don't live in from of their computers, I'm thinking of moving on. I love my Sony RX100, and RX1 would be an option.

Milich or RRS for L plate and grip? Any preferences?

And I spent yesterday evening playing with a bunch of RAW files in SPP and LR4 and thoroughly enjoyed the experience, both the files and the software certainly have their quirks but for me, learning them and working with or around them is part of the fun. As as for the end result....

Luis Cunha wrote:
dakel wrote:
Luis,
You could try a Sony Nex 7, with the metabones speedbooster adapter that allows you to use your Canon lenses on the Sony e-mount.
Good luck, Derrick

I once saw a guy with my macro lens mounted in a Sony Nex-7; I´m not totally sure. But it´s a small sensor with a FF lens. I´m not in the mood for another bayer sensor with A-A just for macro.



Consider a Fuji X-E1 or Fuji X-Pro 1.

They don't have AA filters.

This might be heresy... but life is short. I own the DP2M and the X-E1 and 35/1.4 lens.

I have shot the same images from the same vantage points over the last two weeks, with a pano. setup I described elsewhere (light portable rig). Three-stitch panos, portrait orientation, indexing head on tripod. Grad filter to bring the sky down (though with C1Pro's new gradient tool, that might become a thing of the past, too). Both camera were set to MF, ƒ8, focus at about the 200m distance, manual exposure. Same scene, but different days (and light), but all I was looking for was the detail at infinity (mountains, houses, a lake, etc.).

Now, for the first time today, I compared the Sigma and the Fuji files, both Raw > SPP > TIFF > PTGui > TIFF output for the Sigma, and two paths for the Fuji: Raws developed via either C1Pro 7 or AccuRaw > TIFF > PTGui > TIFF panorama.

Then, just for fun, a final Fuji workflow: using only the Normal jpegs SOOC > PTGui > high quality jpeg output panorama. Smaller files all the way through, but same final pixel count, 10,000-11,000px on the long edge from memory. I did no sharpening on the Fuji files, but the jpeg versions are sharpened lightly in camera as part of the cooking process.

Here's the heresy: I was hard pressed to see any significant difference at the pixel level on a calibrated screen between the Sigma and Fuji images via the Raw workflows. That was a surprise—but when I compared the panorama made with the SOOC jpegs with the Raw workflow TIFFS (massive files) I was amazed to see how good the jpeg-only versions were. At the pixel level, all versions looked excellent (but different, as we have been discussing here) but for me, anyway, I am now wondering if the Foveon advantage is really an advantage. The X-E1 does not have an AA filter, as we all know, so its unsharpened files look very good, too. And the Fuji JPEGs do allow quite a bit of tweaking in Aperture via the Enhancement tools, if you need it.

All around, the Fuji is the easier camera to use, and as I have presbyopia, looking at the LCDs on the backs is my least preferred viewing option (have to put glasses on; etc.). The EVF on the Fuji is excellent (for me) with its adjustable diopter (the reason I did not get the X-Pro 1 when they were released).

Finally, even if the Foveon produces slightly sharper pixel-level detail, there is no more information to be seen there. Roof tiles are clearly seen in all images; objects five miles in the distance can be seen clearly on zooming in—I feel now that I just do not need more resolution. At 50% viewing, which most these days seem to regard as a reasonable proxy for what a print will look like, both cameras produce excellent, but different renditions of the same scene.

Well, that's what I've found, anyway. I decided not to post images because they are different, but my key point is that both cameras produce images that are truly excellent—but the colour rendering is very different. And I would say that a good PP person could take either file and make an excellent print from it.

One final point: after a certain frequency, surely the characteristics of the printer, and how it interacts with the pixel detail, must become significant too, in the final part of the workflow. By this, I mean past a certain resolution, the printer's qualities may make a bigger contribution to how the final image looks, and more resolution may not improve that aspect.

Yes, different renditions, that is the key. We can discuss the 'facts' endlessly and determine ( or not) any number of numerical variables and all very interesting it is too. Heck, the 'Foveon look' may even be a 'cheap trick' as someone said above but does any of this matter to me? Well no, because I just love that Foveon look, and have never been able to get close to it with PP on my CR2 RAWs from the 5D3. But then I'm one of those people who actually like tonemapped images, lol. And it's not just me, everybody, photographers and non photographers alike who I have shown the DP2M images to are astonished, the reactions are overwhelmingly positive. AFAIAC this camera produces photographs with a unique look that I love. It's quite simply amazing.

There's only one thing to do. To the stake! NOW!!!




Happy shooting,
Yakim.

On my way!