Dusk Photos

I took a few dusk photos the other day.  I don’t know if my camera’s lens/CCD is really dirty or if I’m just less tolerant than I used to be but lately I’ve been really disappointed with all my sky photos, because of how grainy they are.  I see an amazing sky and jump in the car to go to my low horizon spot and take a bunch of photos and think "these are gonna be beautiful" and then I come home and hate them all.

Anyway... this set contains 11 of the 81 shots I took, and even these I don’t really like.  If only my camera would go down to ISO 50... or have noise-reduction circuitry...

Posted by Anthony on 25 replies

Comments:

01. Apr 30, 2005 at 01:21pm by Nate:

If your CCD was dirty, you’d know it.  Also, at f/5.6 you don’t see that much dirt on the sensor anyways.  You should see the dirt blobs on mine at f/32.  Anyways, you’re just near the noise floor of the sensor, especially in the dark parts of those images.  Have you ever ran them through a noise removal program?  I ran one of your images through Helicon denoiser.  I also did a 3 point S curve adjustment for a little added contrast because it was a little grayish.  I don’t know why you hate them so much.  They’re not that bad.

http://www.personal.psu.edu/users/n/m/nmb167/ForA/dsc08983_hnf_100.jpg

02. May 1, 2005 at 08:08pm by Anthony:

f/32, I wish.  My smallest f-stop is f/8.

I’m not sure what you mean when you say I’m just near the noise floor of the sensor in the dark regions?

I’ve never tried a dedicated noise removal program, but I’ve tried a few of GIMP’s noise removal filters.  I’ve never thought the added softness/blurriness was worth the noise removal though.  But I don’t think there’s really any way around that tradeoff during post-processing, which is why I really wish my camera went lower in terms of ISO.  The version you posted looks pretty good but it’s a lot darker than the original -- can you post one without the curves adjustment, just to get a straight noise comparison?

03. May 1, 2005 at 09:35pm by Nate:

I uploaded the unedited version to the same address.  Check it out.  Noise removal got a bad rap from companies and people who just didn’t know when to stop.  Gimp never did a great job at it when I was using it.  Then again, I never liked Gimp so I never played with it that much.  This image was run through with the suggested level of noise removal decided by Helicon Noise Removal.  I got a free copy back when the Ukranianes were giving it away.  Now it’s gonna cost ya.

What I meant by the noise floor was just that you have a signal to noise ratio here that isn’t favorable.  Your signal is competing with the inherent thermal noise of the sensor, and apparently losing at the case may be.  When your signal is pretty low you run into these problems. 

Most CCDs are designed around ISO 100 or there abouts.  They offer their best noise characteristics there.  The ones that go down to 50 or lower generally do so at the sacrifice of dynamic range.

Check out some dedicated noise removal software that does the job well and can do both chroma and color noise reduction.  Done right it can really help things without killing your sharpness / quality.

04. May 1, 2005 at 10:24pm by Anthony:

Quoting Nate:

Most CCDs are designed around ISO 100 or there abouts.  They offer their best noise characteristics there.  The ones that go down to 50 or lower generally do so at the sacrifice of dynamic range.

I can set mine to 100, 200, 400. or auto, and auto is nearly always synonymous with 100 in my experience.  I see what you mean about the range-reduction tradeoff, but for a photo like this that doesn’t have a ton of dynamic range to begin with (nothing extremely bright nor dark in it), I’d accept the tradeoff, especially since I could shoot at 100 or "auto" the rest of the time anyway.

I’ll have to look into dedicated denoise software.  I also haven’t spent too too much time trying GIMP’s various plugins for it, but looking at your second version (without the curves adj) I think I could get pretty close to that with something in GIMP.  I’ll try it and let you know.

And for what it’s worth, GIMP has come a long way in the past 2 years or so.  It used to be pretty terrible, but now I use it almost daily and have few complaints.

05. May 2, 2005 at 12:33pm by Nate:

I’m hoping and waiting for a day that GIMP catches up.  PS 7 is woefully outdated for what I usually want to do as well. 

Has GIMP finally added support for 16 bit color channels yet? 

As wonderful as I consider Photoshop to be, Adobe has been annoying the heck out of me with CS.  I use CS more than 40 hours a week at work.  It annoys me that I can’t open tifs as fast with CS as I can with PS7.  I’m talking 10 times slower.  What the heck is that about?!  Could just be that I use a MAC at work...

06. May 2, 2005 at 01:30pm by Anthony:

If you do 40 hours per week of image editing then I’m sure you know a heck of a lot more than I do about where the various apps are at.  I don’t think GIMP supports 16 bit color channels, but there is a fork of GIMP called CinePaint that apparently does.  But man... what kind of stuff are you doing that requires that much resolution?  Are you using an HDTV for a monitor?

07. May 2, 2005 at 02:30pm by Nate:

Man, it’s not about pixel resolution, it’s about utilizing the 12 bit color that comes off the camera sensor.  I’d rather preserve it accurately than down sample it to 8 bit.  It provides a smoother tonal range having a greater number of "steps" to work with.  This becomes much more important the more you get into heavy editing and stuff.

What color setting do you set your monitor to?  256 colors?  16 bit or 32 bit color?  It’s really the same idea.

08. May 2, 2005 at 02:58pm by Anthony:

I have my system’s color depth set as high as possible, which I think is 24bpp.  But that’s exactly my point -- 16 bits/channel gives you 48 bits, much more color resolution than PC monitors are capable of displaying, unless your monitor is an HDTV.

When you say "the 12 bit color that comes off the camera sensor," I assume that only applies to shots taken in RAW format, right?  My camera doesn’t support capturing to RAW, and honestly for the sheer volume of shots that I take (in a given outing at least), I’m not sure if RAW would be as much of a help as a hindrance to me. ... Then again, I guess it depends entirely on the situation and subject.  For my dusk set here, I can see how it would have been advantageous.  But for taking lots of shots of people or events, the amount of post-processing time and effort would definitely cause me to just shoot in JPEG instead.

09. May 2, 2005 at 04:40pm by Nate:

Ya, RAW definately takes some extra time. 

Regardless of how much color depth a given monitor can display, it makes a big difference in editing.  A simple example that stresses my reasoning behind higher color depth:

You take a shot of an overcast sky and convert to gray scale.  So you’ve got 256 potential levels of gray in this image.  If you look at levels, you should have a large peak in the center of the histogram with nothing to the left (shadows) and right (highlights.)  Stretching that histogram (moving the sliders from the far right and far left inwards towards the central peak in the histogram)  creates fewer potential levels of gray.  Maybe you’ve got 120 potential levels of gray now.  After you do levels like that, and you look at the new stretched histogram you’ll notice lots of "holes" or jagged regions of the histogram where there is no longer any data.  This stretched histogram no longer has 256 different levels of gray and it suddenly becomes much easier to see that there are no longer fine gradations between each level of gray. 

Working at 16 bits with the same image gives you the same effect but to a much smaller degree.  256 levels down to 120 vs. 65,000 down to 30,600 levels.

As you said A, it doesn’t matter for a normal monitor.  You probably won’t notice it unless you’re doing editing that’s specifically utilizing the extra steps or color resolution available.  Printing or viewing a final product can be easily done with 8 bit channels, but some editing is really best done in 16.

After all, why keep uncompressed (or in your case, flac’d) waves on your computer?  Because we care about the extra resolution. 

Is this thread OT or what?  haha, it’s been fun.

By the way, I usually never remove noise from images, but I do ADD noise rather regularly. 

Later

10. May 4, 2005 at 01:36am by Anthony:

Quoting Nate:

As you said A, it doesn’t matter for a normal monitor.  You probably won’t notice it unless you’re doing editing that’s specifically utilizing the extra steps or color resolution available.  Printing or viewing a final product can be easily done with 8 bit channels, but some editing is really best done in 16.

I see your point, but since we don’t have any hardware capable of displaying the color information in the extra bits, the only benefit of having it is for posterity -- for one day when we have HDTV monitors -- or if knowing you’re throwing out the data prevents you from sleeping at night.  I’ve seen your stereo (or more to the point: I’ve lifted it), and I know how you feel about these kinds of things, so I understand that this is more of an issue for you than for me :)

Quoting Nate:

After all, why keep uncompressed (or in your case, flac’d) waves on your computer?  Because we care about the extra resolution.

But the difference is that we DO have audio equipment with enough fidelity to display (audibly) the data in WAVs that isn’t there in MP3s in at least some cases.  In a small percentage of all MP3s I’ve encoded I could tell a difference between the MP3 and the original WAV; but on display hardware that’s limited to 32 bpp, a 48 bpp image is identical a 32 bpp image.

11. May 4, 2005 at 03:40am by Steve:

I’m just curious... for which tasks is photoshop 7.0 "woefully outdated"?

12. May 4, 2005 at 07:24am by Nate:

I guess you never noticed the lack of layers, channels, or filters when working in 16 bit?

13. May 4, 2005 at 01:44pm by Steve:

I guess you never noticed the lack of an option for 48 bit color depth in your display properties?  Oh, my fault, you’re on a mac... your "monitors extension" or "system preferences" or whatever the hell they have.

In any event, what program do you use which supports such things at sixteen bits per channel?

14. May 4, 2005 at 09:45pm by Nate:

Your fault Steve, that you didn’t read the rest of the thread to understand that 16 bit is especially relevant for EDITING and not viewing. 

A, it’s not at all about displaying the image on a monitor that I wanted the 16 bit support for.  It’s about what it allows me to do while editing.

As for whether or not we have audio equipment that allows us to hear the true limitations of uncompressed wave.... well, I keep trading up and it keeps getting better.  But that’s IMHO.

Later

15. May 4, 2005 at 10:26pm by Steve:

Your fault Nathan, telling me that I didn’t read the rest of the thread when in fact I did.

The point is, obviously, how are 16 bit channels relevant to ANYTHING when you can’t view the color depth?  You tell me what it allows you to do while editing.

16. May 5, 2005 at 09:20am by Nate:

I know you won’t believe a single thing I say so:

http://www.earthboundlight.com/phototips/8bit-versus-16bit-difference.html

Oh ya, since when are YOU an expert on PS that you feel so compelled to argue this?

17. May 5, 2005 at 10:36pm by Steve:

What you said earlier in the thread is basically that article in summarized form.  But...

Quoting the article::

the extra values represent additional accuracy, rather than additional range. Think of them as being fractional values: we now have 255 additional values between each of our 8-bit values.

On a monitor that can’t display the extra fractional accuracy, the most this can get you is a point of difference in color value in any channel, hardly noticeable.  The only time it’ll really help is when an image is particularly underexposed, and even then you can get the same results in 8 bit with some skillful manipulation.  If you don’t know how to do anything except auto levels, then yea, 16 bits.

I’m not sure why I have to be a photoshop expert to comment on your misguided posts about color depth... my major, however, is basically computer graphics.

18. May 6, 2005 at 07:45am by Nate:

Well it’s pretty obvious that 8-bit and 16-bit have the same dynamic range.  I assumed most people knew that.  The thing I don’t understand is why you don’t understand that greater resolution (additional accuracy) is important when you’re constantly truncating values over and over with each editing step that you take.  I gave an underexposed example because it was SIMPLE FOR YOU TO UNDERSTAND.  It’s not the end all be all reason of the professional imaging world moving to 16 bit resolution.

For the last time, I’ll restate what every article online will tell you.  You introduce color banding and posterization through gamma tweaks, levels, curves, or whatever.  Having a 16 bit working file preserves greater image integrity while applying image adjustments.  Everyone uses a final 8-bit output file for viewing or printing because print drivers don’t support it and it doesn’t matter anyways to the naked eye.

I don’t have to prove anything to you if you choose not to believe the industry.  I did my research and tried it out and found what works best for me.  It also happens to coincide with what the industry believes to be the best approach.  Oh my, what a suprise.

If you have such a problem with this, email Adobe.  Maybe you’ll believe them?

Just for fun, a slightly different article on 16 bit editing.  No suprises here: same conclusions, more examples.

http://www.creativepro.com/story/feature/7627.html

19. May 6, 2005 at 04:39pm by Steve:

These articles you post deal MOSTLY WITH THEORY, not much practicality.  If, when editing a photograph, you are "constantly truncating values over and over with each editing step that you take", THEN YOU PROBABLY aren’t very good at editing photographs. 

You gave the underexposed example because it’s the same one your first article gives.  I cited it because it’s one of the very few instances where 16 bit channels provides even an arguable advantage. 

I’ll restate what I told you.  You can get the same results using an 8 bit image simultaneously with a better-than-amateur AMOUNT OF skill.

It’s not even a matter of belief.  I understand what you’re saying and I understand the theory.  It’s a matter of disagreemeNT.  I can do the same thing with an 8 bit file that you can do with a 16 bit file, and probably faster.

20. May 7, 2005 at 02:49am by Nate:

Ok, so I take a generally well exposed photo and apply a slight level adjustment by bringing it in about 4 levels on the top and bottom, followed by a moderately small S curve to enhance contrast.  When I take a look at the histogram I notice that the stretching of the levels in the 1st step coupled with the stretching of the midband towards the shadows and highlights as a product of the S curve results in EMPTY gaps in the midband of at least 3 shades of gray that are just LOST.  I don’t want holes in my histogram that are 3, 4, 6 or however many levels wide.  Especially in large smooth gradiants like sky where banding is going to become exceedingly obvious.  If I work in 16 bit for just a second to do levels and curves, there are no problems....  It’s so simple. 

Now, care to elaborate how someone making a very small levels adjustment and a small S curve adjustment PROBABLY isn’t very good at editing photographs?  Are you implying that someone making two small adjustments that result in two truncations must be pretty bad at editing? 

Now, explain how your better-than-amateur AMOUNT OF skill bypasses the banding problem without going to 16 bit and all the while doing it faster, since you seem to think it’s possible.  I’ve seen people do it, but not faster, and not necessarily better.

So you’re fine with the theory, but you disagree.  Ok, someone is bound to, and if it’s going to be anyone, it’s going to be you.  So back it up.  How do you deal with banding for example? 

Not much practicality to 16 bit?  I don’t see much practicality in printing a landscape with a lot of banding in the sky, but I do see practicality in working in 16 bit so I can print an 8 bit sky that doesn’t have banding.  So, 16 bit isn’t practical or YOUR eyes can’t see banding? 

I don’t care if you work in 8 bit.  Whatever suites you.  But to make the claim that you can do the same thing with 8 bits and probably faster... well, I’m waiting to see it.  So, where is it?

21. May 7, 2005 at 06:17am by Nate:

"It’s not even a matter of belief.  I understand what you’re saying and I understand the theory."

Ok, so why comment on my "misguided posts about color depth."  I’m STILL waiting for an answer to that one.  If this is preference, practicality, a matter of taste, or a disagreement on any of the previously mentioned aspects, why have you still not given a definitive answer on how my misguided color depths are so misguided?

22. May 7, 2005 at 02:03pm by Steve:

Why are you so capital happy?  Anyway, I’m not gonna go to the trouble of quoting your post again because I don’t think you’re even attempting to follow along, but I’ll answer your questions on a paragraph by paragraph basis.

Post 1:
What I meant by probably not very good at editing photographs was that such a person probably hit a few websites for tips and not much more.  Believe it or not, the S curve is not the end all of photo editing.

My better-than-amateur skill bypasses the banding problem without going to 16 bit and all the while doing it faster, since I seem to think it’s possible, by applying techniques and adjustment layers that you haven’t even mentioned.

With gradient fill layers, hue and saturation adjustments, selective color replacement, and channel mixers, for example.

It’s not usually practical, my eyes can see banding, I’m not color blind at all.

Send me a picture that you believe can only be corrected in 16 bit, and I’ll show you.

Post 2:
Because, for one thing, your edited version of A’s photo wasn’t very good.  For another, you posted some crap as though it were absolute truth, and it isn’t.  You’re STILL waiting for an answer because you never asked before now.

23. May 7, 2005 at 04:06pm by Nate:

The many capitals started out as emphasis early on and later became a parody of some of your excessive capitals, not to mention their odd placement sometimes.

Obviously, you don’t need to talk down to me as if I’m a nobody when it comes to photoshop.  I might not be a "computer graphics major," but I do a lot of image editing on my own and for color and B&W print for the year book portraits and candids at Jostens.  Obviously, my job requires certain bits of knowledge and not others; just as my personal editing does.  I don’t know EVERYTHING about PS, but I do know more than the average person.

My entire point was, and still is:  16 bit makes a simple curve adjustment easier with less degradation.  Total 16 bit support allows me to selectively replace colors, channel mix to alpha channels to conveniently create detailed masks for selective sharpening (or whatever else I feel like adjusting selectively) all the while having adjustment layers for levels, curves, color balance, saturation etc. that can all be changed after print proofing (not to mention color space changes) to fine tune a print.  The more steps I add in, the more degradation of the image.  I don’t want that but it’s a fact of life.  So I work in 16 bit to minimize it.  It’s not readily apparent on all prints; I’ve done some great 8 bit edited ones too.  But I’m not making 4x6 prints (why bother?) and when I’m spending as much per print as I do to do 12x18s (and a couple 6x9 proofs) then why not have a workflow that maximizes quality and consistency? Some images need it more than others.  I think we can agree on that?

I just don’t see the point in starting a project by throwing away useful data in converting from 16 to 8.  I don’t see the point in printing a job that has a crippled tonal range because of editing, regardless of how apparent it is; sometimes it is, sometimes it’s not. 

Yes, A’s thing sucked.  I didn’t want to do much to it other than relieve the noise problem with a one click application and up the contrast a tad to drown out some of the more visible noise in the shadows.  It wasn’t something I spent some time on....  I work to much to have time to fiddle with other peoples stuff to a huge degree.  (I feel like I’m spending way to much time on this frankly.)

As for my posting crap as though it were absolute truth...  What did I state as such?  Your first post questioned my statement that photoshop was woefully outdated.  I’ll assume that’s it because it’s the only thing that makes sense.  My answer is:  "for what I usually want to do."  Simple as that.  I don’t see how that’s absolute truth to anyone else since I was obviously speaking in the first person.  PS7 doesn’t do what I want it to do.  Period.

I’m not sending you some photo that can’t be corrected in 8-bit because they all could be.  What would be the point?  If you want to offer some evidence, submit to my request to publish your workflow to get around banding problems or whatever.  You say that your "gradient fill layers, hue and saturation adjustments, selective color replacement, and channel mixers" will enable you to do the same thing faster, but I don’t know how.  Hey, I could learn something.  The thing I want to know:  How could you easily replicate a non-destructive curve adjustment using those techniques you mentioned?  Quickly and easily to boot.

For the last time, because I never mentioned alpha channels doesn’t mean I didn’t even know they existed.  You talk to me like I’m an idiot that thinks S curves are the end all be all. Grow up. Obviously I’m past that level but I’m not going to sit there and spout out every trite thing I know how to do to impress you or whoever else might be reading this.

I think the addition of greater 16 bit support in CS and CS2 was a welcome and needed addition for graphics professionals.  Sorry you don’t feel that way.  Many do, and if your world only revolves around 8 bit, best of luck to you because I fear you’ll be left behind with the way that the software giants seem to be headed.  My other question is:  Why haven’t you gotten over it? 

You don’t see me crying that the CD replaced the LP from a quality perspective.  That’s life, and some thought it’d be perfect sound forever.  Oh well, they were wrong.  Life goes on.

24. May 7, 2005 at 11:33pm by Steve:

I don’t have some unfounded problem with 16 bit.  I just wanted to add to this thread the (lacking) perspective that you can do the same things with 8 bit that you can do with 16 bit, using different methods.  16 bit and 8 bit images have the same viewable potential, in the end.

Ahaha, I thought it was obvious that my odd capital placement was in parody of your ridiculous capitalization, I guess you missed it.

There’s nothing obvious about my lack of need to talk down to you when it comes to photoshop; you weren’t displaying much evidence of knowledge, how was I supposed to know your skill level.  I was simply seeing your bet and raising:

Quoting Nathan:

Oh ya, since when are YOU an expert on PS that you feel so compelled to argue this?

This conversation is over for that same reason.  You talk down to me, which I expect from someone like you, but then get pissy when I do the same.  And then...

Quoting Nathan, several times:

I guess you never noticed the lack of layers, channels, or filters when working in 16 bit? ... Your fault Steve, that you didn’t read the rest of the thread to understand that 16 bit is especially relevant for EDITING and not viewing. ... Oh ya, since when are YOU an expert on PS that you feel so compelled to argue this? ... I don’t have to prove anything to you if you choose not to believe the industry.  I did my research and tried it out and found what works best for me.  It also happens to coincide with what the industry believes to be the best approach.  Oh my, what a suprise [sic]. ... or YOUR eyes can’t see banding?

...

Quoting Nathan after making the previous comments:

You talk to me like I’m an idiot that thinks S curves are the end all be all. Grow up.

Give me a break.

25. May 8, 2005 at 03:31am by Nate:

Ya, give ME a break.  Talking to you is the opposite of progress.  Always was, always will be.

Reply to this message here:

Your name
Email
Website (optional)
Subject
search posts:

HomeCreate PostArchivesLoginCMS by Encodable