Lawrences SD10 samples

[Laurence Matson]

Tim decided to start the discussion over here in a new thread, so I feed the fires with some more images for you daws to peck at.

Hello Karl. Be polite now. None of your usual stuff. And don't forget the list. Actually, forget the list.

http://www.pbase.com/lmatson/sigma_sd10

Have at it.

--
Laurence Φ€ 08 LL

http://www.pbase.com/lmatson/sd9_images
http://www.pbase.com/sigmasd9/root
http://www.pbase.com/cameras/sigma/sd9
http://www.beachbriss.com (eternal test site)

 

[Laurence Matson]

In a remarkable composite image, this pack is clearly showing its opinion of what it thinks about Karl's criticism while polishing off the remainders of its last hounded victim.



--
Laurence Φ€ 08 LL

http://www.pbase.com/lmatson/sd9_images
http://www.pbase.com/sigmasd9/root
http://www.pbase.com/cameras/sigma/sd9
http://www.beachbriss.com (eternal test site)

 

[Just Looking]

In a remarkable composite image, this pack is clearly showing its
opinion of what it thinks about Karl's criticism while polishing
off the remainders of its last hounded victim.

What's the fourth one named? I only recognize Fear, Uncertainty, and Doubt.

j

 

[Laurence Matson]

Don't know, but I picked him up in Trollhatten when I was in Sweden.

Nomen est omen, I guess.

Carpe diem, which is not a fish.

In a remarkable composite image, this pack is clearly showing its
opinion of what it thinks about Karl's criticism while polishing
off the remainders of its last hounded victim.

What's the fourth one named? I only recognize Fear, Uncertainty,
and Doubt.

j

--
Laurence Φ€ 08 LL

http://www.pbase.com/lmatson/sd9_images
http://www.pbase.com/sigmasd9/root
http://www.pbase.com/cameras/sigma/sd9
http://www.beachbriss.com (eternal test site)

 

[Mike Morrison]

Click on 200% on the viewbar of the original, it has plenty of jaggies. Then compare it to the SPP 2x. I do not have Rick's raw file, maybe he can do the same example for you.

Mike

Something like this:
http://www.pbase.com/image/22805358

Done out at SPP double size. This images tons of jaggies...

http://forums.dpreview.com/forums/read.asp?forum=1027&message=6546598

It has an example of a SPP double sample compared to a regular X3f
file.

Mike

Care to provide an X3f file with jaggies so I can see how the
double size output smooths things for myself.

Now the claim that inkjets aren't good enough and we need a
lightjet to mask the jaggies. That adds a lot to the price of the
system.

Peter

If you have a resize method that removes the jaggies without
trashing the image, please share.

Pete

I've found that if you use the double size output option in SPP
everything is a bit smoother. Furthermore, printing at 300dpi on a
Lightjet (not an inkjet printer) does not proudce jaggies. Perhaps
because of the true continuous tone nature of a LJ print this
doesn't happen.

I'm not saying this eliminates or removes all aliasing that is
caused by undersampling of curved surfaces, especially if you're
talking about large prints. But so far I've yet to see jagged
aliasing in my prints. In the prints of others, depending on how
much jagginess there is, this can easily be manually removed with
the brush tool in PS in a few minutes. IF it's acceptable to clone
out an incidental power line in a photo or use Quantum Mechanics of
PS CS to remove a bit of CA there's nothing wrong with a little
manual brush work here and there if you find a bit of aliasing in a
curved surface.

--
http://www.trytel.com/~pguidry/vacation.html

--
http://www.trytel.com/~pguidry/vacation.html

 

[Peter G]

Kidding right. Obviously if you do a nearest neighbor upsize (ie viewbar 200%) on anything it will INTRODUCE tons of Jaggies.

I am talking about the undersampling jaggies on slanted images that exist in the the Sigma images and how to deal with them.

Peter

Mike

Something like this:
http://www.pbase.com/image/22805358

Done out at SPP double size. This images tons of jaggies...

http://forums.dpreview.com/forums/read.asp?forum=1027&message=6546598

It has an example of a SPP double sample compared to a regular X3f
file.

Mike

Care to provide an X3f file with jaggies so I can see how the
double size output smooths things for myself.

Now the claim that inkjets aren't good enough and we need a
lightjet to mask the jaggies. That adds a lot to the price of the
system.

Peter

If you have a resize method that removes the jaggies without
trashing the image, please share.

Pete

I've found that if you use the double size output option in SPP
everything is a bit smoother. Furthermore, printing at 300dpi on a
Lightjet (not an inkjet printer) does not proudce jaggies. Perhaps
because of the true continuous tone nature of a LJ print this
doesn't happen.

I'm not saying this eliminates or removes all aliasing that is
caused by undersampling of curved surfaces, especially if you're
talking about large prints. But so far I've yet to see jagged
aliasing in my prints. In the prints of others, depending on how
much jagginess there is, this can easily be manually removed with
the brush tool in PS in a few minutes. IF it's acceptable to clone
out an incidental power line in a photo or use Quantum Mechanics of
PS CS to remove a bit of CA there's nothing wrong with a little
manual brush work here and there if you find a bit of aliasing in a
curved surface.

--
http://www.trytel.com/~pguidry/vacation.html

--
http://www.trytel.com/~pguidry/vacation.html

--
http://www.trytel.com/~pguidry/vacation.html

 

[Mike Morrison]

Sure, anytime you zoom in on a sharp image your going to see alaising. That is not what I was responding to. You aked for an example of a raw X3f file that was double sampled in SPP. I provided that. Double sample viewed at 100% would be the original at 200%. In other words, the alaising is much less noticable with the double sample when viewed at 100%, zoom that to 200% and you got jaggies.

Seems to me that anti-alaising via software can resolve the issues you are concerned about. It does for me.

Mike

I am talking about the undersampling jaggies on slanted images that
exist in the the Sigma images and how to deal with them.

Peter

Mike

Something like this:
http://www.pbase.com/image/22805358

Done out at SPP double size. This images tons of jaggies...

http://forums.dpreview.com/forums/read.asp?forum=1027&message=6546598

It has an example of a SPP double sample compared to a regular X3f
file.

Mike

Care to provide an X3f file with jaggies so I can see how the
double size output smooths things for myself.

Now the claim that inkjets aren't good enough and we need a
lightjet to mask the jaggies. That adds a lot to the price of the
system.

Peter

If you have a resize method that removes the jaggies without
trashing the image, please share.

Pete

I've found that if you use the double size output option in SPP
everything is a bit smoother. Furthermore, printing at 300dpi on a
Lightjet (not an inkjet printer) does not proudce jaggies. Perhaps
because of the true continuous tone nature of a LJ print this
doesn't happen.

I'm not saying this eliminates or removes all aliasing that is
caused by undersampling of curved surfaces, especially if you're
talking about large prints. But so far I've yet to see jagged
aliasing in my prints. In the prints of others, depending on how
much jagginess there is, this can easily be manually removed with
the brush tool in PS in a few minutes. IF it's acceptable to clone
out an incidental power line in a photo or use Quantum Mechanics of
PS CS to remove a bit of CA there's nothing wrong with a little
manual brush work here and there if you find a bit of aliasing in a
curved surface.

--
http://www.trytel.com/~pguidry/vacation.html

--
http://www.trytel.com/~pguidry/vacation.html

--
http://www.trytel.com/~pguidry/vacation.html

 

[Just Looking]

Laurence, these wires I cropped from your "street before" image have a jaggie look, almost like twisted rope. Must be aliasing... ;^)
(or was http://www.pbase.com/image/23115153 ; somebody else's shot?)

Just Pecking

Tim decided to start the discussion over here in a new thread, so I
feed the fires with some more images for you daws to peck at.

 

[Laurence Matson]

JP or JL,

This is another image illustrating how low-tech rural communities can be, and it presents one of the great challenges - it is also sometimes referred to as the last mile syndrome - facing the digital imaging industry. Boy, I hope Foveon can solve this one two, but it don't look too good.

You see, JP or JL, in olden days before "cellular" phone came along, they used to have fixed-line telephony. And primitive forms of that technology involved double-lead wires that were TWISTED*.

The challenge facing the industry is correcting this wired situation through better all-log-rhythms.

In this case, the VERY SHARP Foveon technology sensor (which camera did he use, by the way?) is a real problem. I am sure the use of an anti-alien filter could handle this in a snap.

As to whose image you took. Well it must be mine, since my name is found in these threads.

Looks pretty good for a sports shot, don't you think? I cropped out a section below to show how good.

• all caps in compliance with the FACTS style rule; see Skunk and Wagalls, 3rd.

--
Laurence Φ€ 08 LL

http://www.pbase.com/lmatson/sd9_images
http://www.pbase.com/sigmasd9/root
http://www.pbase.com/cameras/sigma/sd9
http://www.beachbriss.com (eternal test site)

 

[Just Looking]

You see, JP or JL, in olden days before "cellular" phone came
along, they used to have fixed-line telephony. And primitive forms
of that technology involved double-lead wires that were TWISTED*.
...

• all caps in compliance with the FACTS style rule; see Skunk and

Wagalls, 3rd.

L, yes, it's a very current topic; electrifying with a twist; with lots of potential. Can't take the high tension? Feeling wired? Then get grounded. Telegraph this:

If you're going to counter my observations with FACTS*, you need to get your facts straight, instead of just "making up stuff" like some people we know. The best way to get your facts straight is to just close your eyes and "google". I found out this:

Service Drop -- A SERVICE DROP is the lines running from a utility pole to a customer’s house. Usually, a service drop is made up of two 120-volt lines and a neutral line, from which the customer can obtain either 120 or 240 volts of power. When these lines are insulated and twisted together, they are referred to as a triplex cable.
http://www.duquesnelight.com/understandingelectricityupdate/electricterms.html

so you see, you made up that "double-lead" tele-phoney nonsense. It's a triplex...with the neutral not insulated. Now, don't deny that it looks ropey.

j

 

[Laurence Matson]

JL or JP (you never cleared that up),

Well how 'bout that. Well, it depends on how you phase it, I guess. That "google" thing is sure good, that's for sure.

Must be a California thingie, since Mr. Oogle himself starts governing today. Sacremento never had it so good. (How long will that stay the state capital with the new recall effort underway; I hear Mr. Oogle wants to move it to Silicon Valley to honor all of the ladies who have undergone cosmetic surgery in certain places if you catch my "continental" drift - boy, talk about a technology gap!) - but I digress.

So how was I supposed to know about three-wire telephone lines? And since when did telephones work on both 120 and 240? Boy, you sure got your FACTS* wrong.

But one fact you cannot ignore is that this seems like a solvable problem for the X3 sensor (catch the relationship in that transition?). With three sensors per location, it should easily be able to handle those TWISTED WIRE* jaggies before all those over-powered handsets blow up in rural America.

Boy, I hope Mr. Oogle has this right at the top of his agenda, 'cause I just found out this was not my image but one taken by someone else in rural CALIFORNIA and I could get in trouble for defamation of something. Maybe you better check your attribution rules.

Seems like we're all getting on edge around here. Maybe Mr. Oogle should pass a resolution on this too. Just don't start trying to explain it to him.

• all caps in compliance with the FACTS style rule; see Skunk and Wagalls, 3rd.

--
Laurence Φ€ 08 LL

http://www.pbase.com/lmatson/sd9_images
http://www.pbase.com/sigmasd9/root
http://www.pbase.com/cameras/sigma/sd9
http://www.beachbriss.com (eternal test site)

 

[mirko b]

Sorry for the infiltration in your discussions...

I'm a reader of many of the forums here on dpreview beause I'm intrested in buying a digital camera, but I really can't understand some of the guys that post their opinion with the only ends to but some camera brand in shame.

I think that if someone owns a Nikon, he (or she) will surely be happy with it, and the same for those who own Canon, Sigma, Olympus etc..., it's obvious (it's hard for someone to admit "I have made the wrong choiche")

Every digital camera owner have a mission writing in all the forums: for someone it is to search for help, for someone else it is to make comparisons between different cameras, ...

But for someone it seems that the only mission is to put in shame someone who posted their pics or their opinion, or a camera maker. I really can't understand them.

I've foud very intresting and pacific discussions here, and some censurable ones (that goes nowhere, and fortunately are just a few).

Is it really so hard to be tolerable with those who have different opinions?

I hope someone listen to my advice.

For those who don't, I hope that maybe they will grow sometime..

(excuse me for my poor english)

 

[ChromeLight]

Your assertions are somewhat difficult to test.

Care to provide an X3f file with jaggies so I can see how the
double size output smooths things for myself.

Now the claim that inkjets aren't good enough and we need a
lightjet to mask the jaggies. That adds a lot to the price of the
system.

Peter

I'm not sure I have any with aliasing, since it happens so rarely, and I don't shoot architecture. (That's best left to a view camera.) But I'll see, when I get the time. You can actually see a comparison between the double output of SPP vs. bicubic, SI bicubic, and genuine fractals if you look at the posts of Carl Rytterfalk in the Sigma forum. This was done many months ago. I'm sorry I don't have the URL to the discussion. You can see there for yourself how the double output option is smoother than bicubic or GF. Some don't like it because it is too smooth. Furthermore, S-Spline has an anti-aliasing option in their program, which I assume would work to eliminate any jaggies, if they are present. Also, the new Bicubic smoother option in PS CS might do something similar.

Come to think of it, I'll see this week if I can post some 100% crops from double size, bicubic and bicubic smoother. That's the best I can do on a short term basis.

 

[Michael Long]

But for someone it seems that the only mission is to put in shame
someone who posted their pics or their opinion, or a camera maker.
I really can't understand them.

Unfortunately, no product is prefect. Every product has flaws. And no single product is best suited for every possible use. And as such, part of what we do here is look past the marketing hype and expose those flaws, discuss them, determine if work-arounds exists, and so on.

I mean, really, how many companies are going to tell you they're NOT the best solution?

But in any group of people who have made the decision to purchase an item, there are those who have a heavy, personal investment in that decision.

Those people will insist their product is superior, that it has no flaws, and will promote it as such to everyone, whether it's suitable or not.

There was a recent post, for example, where someone was looking for a suitable camera for high-end architecture and landscape photography. That type of photography demands high levels of detail, and often requires the use of specialized, professional UWA and tilt-shift lenses.

And of course, someone dropped in promoting the SD9, even though it doesn't really have high spatial resolution, and even though Sigma doesn't really have those lenses in it's lineup.

Foveon is heavy into promoting the superiority of it's concept. Don't get me wrong, it's a great concept. But the current implementation of that concept has constraints that mean its not always the best solution to a given problem.

Since it's "our job" here to go past the hype, if you're going to claim you're the best, you'd better be able to back it up.

 

[Sean Keegan]

Hey Mike,

That doesn't explain the blotches in the sky, in fact I see blotches in many areas.

Regards,
Sean

Some of y'all really go out of your way to make problems where
there are none. You need to out more.

"We're all in this mess together, so let's act like it"

Mike

After printing several of the images, I'm convinced the camera
needs an AA filter. The house pic has a lot of diagonals, with very
strong, very sudden aliasing occuring, which became very obvious in
the prints.

One of the better prints came from running a set of slight gaussian
blurs on the image, then resharpening. Although the CA caused by
the sigma lenses did make that correction a bit difficult.

 

[Sean Keegan]

Hey Michael,

I broke down and decided to waste some ink and print out the ISO 100 sample at 13x19. I agree... very digital looking aliased output. Very obvious even at arms length.

Regards,
Sean

After printing several of the images, I'm convinced the camera
needs an AA filter. The house pic has a lot of diagonals, with very
strong, very sudden aliasing occuring, which became very obvious in
the prints.

I'm baffled.

I did a nice 12x18 print of the ISO 400 shot, and must say that
there's only a tiny amount of "jaggieness" visible. Certainly not
enough to be objectionable, but enough to find if that's what
you're looking for.

Maybe you need a better printer, or better way to resize for
printing? What kind of old dot matrix are you running there?

One of the better prints came from running a set of slight gaussian
blurs on the image, then resharpening. Although the CA caused by
the sigma lenses did make that correction a bit difficult.

I hope you didn't strain yourself. Blurring and resharpening does
sound tortuous.

By the way, do you recall where we showed that the SD9 does already
come it at a pretty good compromise with its anti-aliasing? Did
you follow?

j

 

[Sean Keegan]

Wow, That looks horrible.

ISO 1600 looks really bad sized down! The blotchyness of the sky is "inyourface'!!!

I hope you were not trying to use that image as an example of good noise at ISO 1600!

Regards,
Sean

Thanks to Lawrence for the latest SD10 samples.

It's Laurence, with a "U". And we're talking Sigma SD10, not Canon
SD10, in case anyone was lost.

Laurence's posting is here:
http://forums.dpreview.com/forums/read.asp?forum=1027&message=6659505
and images here:
http://www.pbase.com/lmatson/sigma_sd10

Here's an ISO-1600 sample:

Overall, Im
suitably impressed with the cameras level of detail, and performace
at higher ISO's.(esp at 400)

Me too.

Not so impressed with your reactions, though. My comments here:

http://forums.dpreview.com/forums/read.asp?forum=1027&message=6667074

j

 

[Just Looking]

Wow, That looks horrible.

ISO 1600 looks really bad sized down! The blotchyness of the sky is
"inyourface'!!!

I hope you were not trying to use that image as an example of good
noise at ISO 1600!

Regards,
Sean

Sean, I don't recall expressing any value judgements one way or the other, other than agreeing that the level of detail is impressive at ISO 400. It's just an illustration of "what you get at ISO 1600 (as downsized by pbase)" to tease people into following the links to see more, full-size.

I take it from your comment that other cameras can do better at 1600. I wouldn't know, never having seen anyone post full-size originals at ISO 1600. Certainly wouldn't surprise me though, as 1600 is in the Sigma SD10's "extended" mode only, meaning probably that it's not where you should expect to get low-noise pictures.

I just think it's great the Laurence was open enough to share images that show some of the limits of what this new camera can do.

j

Here's an ISO-1600 sample:

Overall, Im
suitably impressed with the cameras level of detail, and performace
at higher ISO's.(esp at 400)

Me too.

Not so impressed with your reactions, though. My comments here:

http://forums.dpreview.com/forums/read.asp?forum=1027&message=6667074

j

 

[Sean Keegan]

I take it from your comment that other cameras can do better at
1600. I wouldn't know, never having seen anyone post full-size
originals at ISO 1600. Certainly wouldn't surprise me though, as
1600 is in the Sigma SD10's "extended" mode only, meaning probably
that it's not where you should expect to get low-noise pictures.

This is what I am discovering from you Just Looking. You really do not know what other cameras are capable of. You just have LOTS of your theories, but no experience. You only see the greatness of the X3 techniques, and cannot see that bayer is actually better in many ways currently. If there was a X3 at the same MP as bayers (6mp) then there would be no arguements that the X3 is the best (in most areas anyway... noise looks like a very weak spot of the X3 and I think it would be worse with 6mp).

In answer to your question, YES I definitely think that at ISO 1600 shrunk down that drastically you should not see ANY evidence of noise at 1600 any longer. Most DSLRs at 1600 shrunk down this far would have no noise visable at all. If you want samples let me know.

I just think it's great the Laurence was open enough to share
images that show some of the limits of what this new camera can do.

Yep me too. But when you do that you open the door to the critics. That was the whole point of his post. I hope it was not to prove how superior the SD10 is. Instead I hope he wanted to learn what the rest of the world thinks about the SD10, no holds barred. People are just being honest here. No reason to be defensive.

Regards,
Sean

 

[Sean Keegan]

However, most of these disagreements comes from those existing
Bayer lovers who tends to be close-mined that keeps stumping on
Foveon's furture possibilies and keeps going to SSTF bashing the
chip and product using unfounded information as facts. After
dealing with that kind of pointless bashing for over 1-1/2 years, I
am sure you wont be too friendly to those people as well,
especially when there are only a handful of those that keeps doing
it.

The problem is exactly this. I doubt anyone here does not see the potential of the X3, but right now, it it NOT better than bayer sensors. Bayer is moving forward (still) while Foveon has stagnated with another 3mp X3 chip. I see no real improvements. Beter noise, but still not as good as bayer. No better resolution. Still the same old aliasing problems. ETC. ETC.

We are just being honest critics. When is Foveon going to stop being "comparable" and start being better? If I was tied to Sigma lenses now after being roped in with the SD9, I would be a little worried that the SD10 is the most progress Foveon can show for itself after the 1-1/2 years you are talking about. Meanwhile in the same timeframe, bayer is now pusing 14 megapixels and even farther as we will see in January. You can keep your 3MP super duper X3 chip.

Yes I am upset at being fooled by the marketing scheme of Sigma saying the SD10 was a 10MP camera, in case you can not tell.

Regards,
Sean

As many of us in SSTF always say ... Quit the bashing and let the
photos do the talking.

We are, they are talking, and they are not any better than any 6mp bayer DSLRs output, and in some cases worse. Open your eyes! Make some prints, the results are obvious! We are not bashing Foveon, we are mearly pointing out that these samples, and most others I have seen, are not shining examples of the supposed 'revolution' we could expect from this technology. I thought it would be a revolution if you would look back at my post history. I was all for the X3 and its technology. Now it appears that Foveon and their revolution was just a flash in the pan. HYPE! That is all it was. Looks like the fuji S3 is in my future, not my dream camera of a 10MP X3 camera. Not that I am complaining, I think the S3 is going to be incredible!! Foveon has a lot of catching up to do.

Regards,
Sean