| Author | Thread |
|
|
06/03/2006 01:26:45 PM · #776 |
Originally posted by stdavidson:
OK... In honor of your suggestion I added a gradient to my panorama...
Goal: Make a better balanced sky yet retain its natural qualities.
Here is the result:
//www.pbase.com/azleader/image/61217163
Post:
...
... |
Very good! Now that you've seen the power of it, it's hard to go back isn't it?
The interesting thing is that "artificial" gradients actually end up looking quite natural if used correctly at the right opacity and blending mode. On skies, I often pick a color from the top of the sky, darken it a bit, and on a new multiply layer apply a color-to-trasparent grad.
However, I even use the grad tool to burn-in my photos. If you do a black-to-trasparent grad in overlay mode, this can surprisingly look quite nice. Such is the versatility of the gradient tool.
Anyway, here's to seeing more gradient-enhanced landscapes in the future ;)
|
|
|
|
06/03/2006 02:14:27 PM · #777 |
Originally posted by justin_hewlett:
Anyway, here's to seeing more gradient-enhanced landscapes in the future ;) |
Another thing I like about gradients which is also what I like about color painting is that it does not add any noise to the image whatsoever like adjustment layers do.
However, that is not to say I've jumped on the bandwagon yet. Here is why...
My main interest in fine art photography is to make and sell large prints. Generally speaking that means upscaling the file sizes significantly for larger 300 dpi prints, like 16 X 20 for example. Any little defect then becomes major when you do that. You can get away with all kinds of things working with tiny 72dpi web graphics that you can't with large prints. Practically every modification to other people's pictures you see in this discussion do that and you can hardly tell, but you'd never get a decent print from them.
Moreso than color painting, gradients introduce the issue of digital banding caused by their smooth tonality shifts. That happens because the limited number of colors supported in 8-bit produces discontinuous color shifts with linear borders that can show up as color banding on large scale prints. This is something you cannot generally see in a web graphic. This can be a particularly big problem because many commercial processors, including DPCPrints, requires 8-bit .jpg print files.
I very specifically looked closely at the sky in my full sized pano at 300% size looking for linear banding that would show up in big prints. It did not look to bad so I am encouraged, but I could see some. I will not be convinced gradients are OK until proven in large prints.
In the final analysis I'm guessing gradients would work well if done on 16-bit mode master files. I've found that when 16-bit, which supports a lot more colors, is downscaled to 8-bit output files that the conversion scatters the discontinuous color shifts making them more diffuse which reduces the banding effect in them.
|
|
|
|
06/03/2006 02:58:56 PM · #778 |
Originally posted by stdavidson:
I would never have figured it all out without fracman's help.
All negatives aside, I'll use Hugin.
|
First, it was my pleasure. Yesterday made a really nice break from the crappy software I have to maintain at work :-)
Now, I just discovered another use for Hugin. Turns out that in addition to doing panos, it can also do barrel and perspective corrections. There are a couple tutorials on the Hugin site Hugin Site. Before recommending it, though, I tried it myself:
Before
After
The tutorials on perspective correction are not bad, although they deal with an earlier version of the software, so some of the screenshots don't line up exactly. If you run into problems, let me know and I'll see if I can help.
BTW, Bear, sorry the picture isn't a landscape, but the strong lines help illustrate the purpose. The technique would work quite well for urban landscapes.
Jon
Message edited by author 2006-11-14 16:24:37. |
|
|
|
06/03/2006 03:54:51 PM · #779 |
Originally posted by stdavidson:
Originally posted by ursula:
Wow, thank you! I think I might know how to do that. How is this?
 |
Very nicely done image. You originally mentioned layer masks but did not say how you actually applied them. So how did you use layer masks? |
[/quote]
I made two versions from RAW, one very light, the other quite contrasty. I stacked them (light on bottom, contrasty above), then made a mask of the second (contrasty layer) and painted over it with a black brush to reveal portions of the underlaying layer. It worked beautifully!
Originally posted by stdavidson:
[quote=ursula]
The sharpness of this image is exceptional on the geyser and pool. That really makes this image special. What sharpening technique did you use to get such crisp sharpness?
I'm still bummed I did not get to meet up with you to take pictures when you were down this way. :( Oh well... next time!
Btw... I really like this picture of the Mintons: 
I know exactly where you took that one. It is one of the places that I'd hoped to take pictures with you guys. I'm jealous that it is better than the pictures I've taken there. It also has perfect sharpening and color.
We have got to take pictures together sometime so I can steal all your great ideas. LOL! |
Thank you. Both were sharpened using the method you described (I think in a different thread, where you create two duplicate layers of the finished image, apply USM to both, blend one in lighten, the other in darken. In the Yellowstone image, I mostly kept the lighten layer (around 80%), darken around 20%. I also erased portions of the image (clouds in particular) that I didn't want sharpened. In the Monument Valley image I did the opposite (darken around 80%, lighten around 20%).
The Monument Valley image is done all wrong, middle of the afternoon, direct sun, no filters - I really, really wished I could have been at Monument Valley in the early morning (or late afternoon).
Message edited by author 2006-06-03 16:05:55. |
|
|
|
06/03/2006 04:03:19 PM · #780 |
Originally posted by Bear_Music:
Originally posted by ursula:
Hi guys. I've been [quietly] following this thread, and I think I finally learned a bit how to work with layer masks, in particular the method to get more details using two different processed images. |
The image looks very nice indeed, and version 3 is the best of the versions, yes. The steam is especially well-rendered. What exactly do you mean by "getting more details using two different processed images"?
Are you referring to the merging of two different RAW exposures (HDR merging) or are you referring to the "contrast masking" we have worked with so much in here? While I know we have mentioned HDR in passing, I don't think we've actually "worked" with it in this thread, unless I'm forgetting something (always a possibility). Can we see an unaltered original on this shot? A jpg straight from RAW?
R. |
I thought I was working with contrast masking as you guys have been talking about in this thread (but sometimes I confuse threads, there are so many). I made two versions from RAW, one very light, the other quite contrasty. I stacked them (light on bottom, contrasty above), then made a mask of the second (contrasty layer) and painted over it with a black brush to reveal portions of the underlaying layer. It worked beautifully! That's contrast masking, isn't it?
Here is a straight JPG from RAW, no adjustments whatsoever (except resize)
I think I'm really going to like this method, but I need to experiment more with it, a lot more. |
|
|
|
06/03/2006 04:16:10 PM · #781 |
Originally posted by ursula:
...
Thank you. Both were sharpened using the method you described (I think in a different thread, where you create two duplicate layers of the finished image, apply USM to both, blend one in lighten, the other in darken. In the Yellowstone image, I mostly kept the lighten layer (around 80%), darken around 20%. I also erased portions of the image (clouds in particular) that I didn't want sharpened. In the Monument Valley image I did the opposite (darken around 80%, lighten around 20%). |
Ah, yes. That is a good technique. The thing you did different from me in the Yellowstone shot is is that you reversed the percentages of light and dark layers. I suspect your USM values are different from what I usually use.
I use Smart Sharpen now which is better than the PS7 USM technique. I still struggle with getting sharpening right. I'm just glad I don't normally save the sharpenning, that way when I finally get it all figured out right I can go back and correct all my images. LOL!!!
Btw, your "contrast mask" method is different from Bear_Music's one where you select and put highlights and shadows on different layers. Your method is more like what Photoshop calls "Merge to HDR..." and called the two exposure "Contrast Mask" here:
//www.shutterfreaks.com/Tips/Two-ExposureContrastMaskingPSE.html
Message edited by author 2006-06-03 16:26:05.
|
|
|
|
06/03/2006 04:22:15 PM · #782 |
| Smart sharpen - is that PS CS2? |
|
|
|
06/03/2006 04:30:48 PM · #783 |
Originally posted by stdavidson:
Btw, your "contrast mask" method is different from Bear_Music's one where you select and put highlights and shadows on different layers. Your method is more like what Photoshop calls "Merge to HDR..." and called the two exposure "Contrast Mask" here:
//www.shutterfreaks.com/Tips/Two-ExposureContrastMaskingPSE.html |
BUMMER! For once I thought I was getting it right. Oh well. |
|
|
|
06/03/2006 04:34:50 PM · #784 |
Originally posted by ursula:
Smart sharpen - is that PS CS2? |
Yes.
|
|
|
|
06/03/2006 04:52:14 PM · #785 |
Originally posted by ursula:
I thought I was working with contrast masking as you guys have been talking about in this thread (but sometimes I confuse threads, there are so many). I made two versions from RAW, one very light, the other quite contrasty. I stacked them (light on bottom, contrasty above), then made a mask of the second (contrasty layer) and painted over it with a black brush to reveal portions of the underlaying layer. It worked beautifully! That's contrast masking, isn't it?
Here is a straight JPG from RAW, no adjustments whatsoever (except resize)
I think I'm really going to like this method, but I need to experiment more with it, a lot more. |
I took the unedited version above and ran it through "my" version of contrast masking. Absolutely no other adjustments made; no color adjustments, no local burn/dodge, no gradients or vignetting, no sharpening, just the contrast masking, and got this:
R.
Edit to add; from opening to save-as, took about 90 seconds...
Message edited by author 2006-06-03 16:55:31.
|
|
|
|
06/03/2006 05:00:10 PM · #786 |
Rob, that's wonderful! I have to re-read the thread once again. The biggest difference I see is that the portion to the right centre doesn't have much detail in your version, but I'm assuming that with a bit of work it would.
Sorry that I got the wrong method. |
|
|
|
06/03/2006 05:04:52 PM · #787 |
Originally posted by ursula:
Originally posted by stdavidson:
Btw, your "contrast mask" method is different from Bear_Music's one where you select and put highlights and shadows on different layers. Your method is more like what Photoshop calls "Merge to HDR..." and called the two exposure "Contrast Mask" here:
//www.shutterfreaks.com/Tips/Two-ExposureContrastMaskingPSE.html |
BUMMER! For once I thought I was getting it right. Oh well. |
Who said you didn't get it right? I know three different "contrast masking" methods and probably each one of them will get you where you wanna go, and that is not counting Photoshop's Shadow/Highlight feature.
Results are what matter and it looks pretty darned good to me. :)
|
|
|
|
06/03/2006 05:07:15 PM · #788 |
Originally posted by ursula:
Rob, that's wonderful! I have to re-read the thread once again. The biggest difference I see is that the portion to the right centre doesn't have much detail in your version, but I'm assuming that with a bit of work it would.
Sorry that I got the wrong method. |
Hey there's nothing to be sorry about; in PS there's a gazillion ways to skin a cat. You're correct, with a little bit more work we could spruce up that version of mine dramatically. I'm cooking a 6-course Indian meal for 8 right now, but I'll take a stab at doing it up right later tonight or sometime tomorrow. (By "right" I just mean "How I'd do it if it were mine"; your mileage may vary, jejejeâ„¢)
R.
Edit to add; what Steve just said, I meant to be implying with "gazillion ways"; all that matters is the end result. Of course, some methods get from point A to point B more efficiently than other methods, so that's something to take into consideration. To whatever degree the cntrl-alt-tilde technique of contrast masking allows you to reduce the amount of "erasing" you have to do, why then so much the better. I've found it to be very efficient most of the time, and I like that in a process :-)
Message edited by author 2006-06-03 17:10:11.
|
|
|
|
06/03/2006 05:10:49 PM · #789 |
Originally posted by Bear_Music:
Originally posted by ursula:
Rob, that's wonderful! I have to re-read the thread once again. The biggest difference I see is that the portion to the right centre doesn't have much detail in your version, but I'm assuming that with a bit of work it would.
Sorry that I got the wrong method. |
Hey there's nothing to be sorry about; in PS there's a gazillion ways to skin a cat. You're correct, with a little bit more work we could spruce up that version of mine dramatically. I'm cooking a 6-course Indian meal for 8 right now, but I'll take a stab at doing it up right later tonight or sometime tomorrow. (By "right" I just mean "How I'd do it if it were mine"; your mileage may vary, jejejeâ„¢)
R. |
Oh, a 6 course Indian meal sounds soooooo good! Yummmmm. |
|
|
|
06/03/2006 05:14:38 PM · #790 |
Originally posted by ursula:
Oh, a 6 course Indian meal sounds soooooo good! Yummmmm. |
jejejeâ„¢ Right now I'm braising minced leg of lamb in fresh orange juice, and a ton of Indian spices. I got a cabbage thing, I got a red kidney bean curry, I got jasmine rice, I got fresh corn kernels all jazzed up, will cook at last second, I got gawd-knows-what... Half the guests are vegetarian. I haven't cooked Indian in probably 6-7 years, except I make "leftover curry" all the time, so I'm dusting off old skills.
The hard part was finding all the herbs and spices I needed. When's the last time you used asafetida?
R.
|
|
|
|
06/03/2006 05:18:03 PM · #791 |
Originally posted by justin_hewlett:
Originally posted by ursula:
Smart sharpen - is that PS CS2? |
Yes. |
Btw... Smart Sharpen is basically the USM method Ursula used to sharpen her two pictures with a little more control.
|
|
|
|
06/03/2006 06:16:49 PM · #792 |
I have only tried out the version included in CS2, which I found to be very crappy, and Panorama Factory. I don't have a special panorama tripod head, so it is a bit more difficult because each merge has to be done manually, but when it works, it works wonderfully.
I have found it to be helpful to check the option to do the exposure correction and brightness fallof for viewing purposes during stitching, but then save it to a .TIF file without those adjustments. Otherwise it completely blows out the highlights for some reason. |
|
|
|
06/03/2006 06:18:28 PM · #793 |
| I do have one question also. What lens do you find works best with your panos? I like the more realistic-looking panos rather than the huge distorted ones, but getting the focal length right can sometimes be a problem also, as well as distance from the subject. |
|
|
|
06/03/2006 07:10:20 PM · #794 |
OK - suggestion for the next lesson.
How about how to shoot landscapes in low light / at night?
|
|
|
|
06/03/2006 07:19:17 PM · #795 |
Originally posted by traquino98:
I do have one question also. What lens do you find works best with your panos? I like the more realistic-looking panos rather than the huge distorted ones, but getting the focal length right can sometimes be a problem also, as well as distance from the subject. |
Welllll... my camera only has one lens so that is what I use. LOL! I take pictures with the camera mounted on a regular tripod without a panorama head. It helps much to have your tripod level, though. ;)
All my panos are landscapes so I usually have the focus set to infinity for each image, the focal length set at one value only for all frames in the pano and the f/stop high for maximum depth of field. I turn off autofocus and have all camera settings exactly the same in manual mode. I make sure the images overlap by at least 1/4th of the frame and use objects within the field of view to decide how to reposition the camera for the next shot.
You are right that Photoshop's panorama builder, Photomerge, in CS2 sucks. I found Panorama Factory produced good looking output but that entering all the merge control points was "cumbersome". LOL!
Scroll back in this discussion and there is a summary of my views of several products I experimented with this past week. Comparative raw output from my experiments with simple panoramas can be seen at the bottom of my panorama page. The software used is hidden in the image title. :)
//www.pbase.com/azleader/panos
Fracman is a great resource of information about panoramas.
Good luck.
|
|
|
|
06/03/2006 07:23:04 PM · #796 |
Originally posted by samchad:
OK - suggestion for the next lesson.
How about how to shoot landscapes in low light / at night? |
Bear_Music may already have something in mind.
We've pretty much beat panoramas to death this week and it is time to move on. LOL!
|
|
|
|
06/04/2006 12:13:52 AM · #797 |
Originally posted by stdavidson:
Originally posted by samchad:
OK - suggestion for the next lesson.
How about how to shoot landscapes in low light / at night? |
Bear_Music may already have something in mind.
We've pretty much beat panoramas to death this week and it is time to move on. LOL! |
I don't know about beating panoramas to death, certainly software for merging panoramas! :)
I have only shot 2 panoramas in my life and have come to the conclusion I know even less about taking them than taking other photographs! I shot the first one last summer and just used the software that came with my R707 to merge the 4 shots together. I shot the second one today and tried to merge the 5 shots manually instead, just for the experience and to see what would happen... (didn't come out very well, maybe next time)...
 
I think I am probably overlapping too much when I am shooting. How much of an overlap between shots do you guys who shoot a lot of panos usually use? Is the fisheye look with the 5 shots because of the way I move my camera? How do I avoid that? What settings on the camera should I be aware of to make sure the colours stay consistent?
I have read the thread and I apologize if these concepts have been covered and I missed reading it somewhere amidst the software discussions, but as someone who is completely new to the concept of panoramas, I would really appreciate some basic set up tips for shooting the photos, as well as all the great info on what to do with them after. :D |
|
|
|
06/04/2006 09:28:13 AM · #798 |
Originally posted by Prism:
I think I am probably overlapping too much when I am shooting. How much of an overlap between shots do you guys who shoot a lot of panos usually use? Is the fisheye look with the 5 shots because of the way I move my camera? How do I avoid that? What settings on the camera should I be aware of to make sure the colours stay consistent?
I have read the thread and I apologize if these concepts have been covered and I missed reading it somewhere amidst the software discussions, but as someone who is completely new to the concept of panoramas, I would really appreciate some basic set up tips for shooting the photos, as well as all the great info on what to do with them after. :D |
Despite what Steve says, I'm not that much of an expert, as I have yet to shoot a pano that didn't require a whole bunch of monkeying around to get somewhat palatable.
Since I don't have a fisheye lens yet, I can't say for sure how many shots is "ideal". What I can say, however, is that if you have to choose between less overlap or more, err on the side of more. My pano of the Mississippi River shot from a dam was really tough to do because I'd only overlapped shots 5 and 6 by 450 pixels (out of 2000 pixels wide for the shot, which is about 22%). Hmm, doing the math, that doesn't seem bad. The problem I had is that the only features that I could use as control points were the tops of trees that were 1-2 miles away. Everything else on that overlap was either rapidly moving water or open sky.
When I'm shooting panos these days, I'm not looking so much for how many pixels or a certain percentage overlap. What I try to look for is that I'm overlapping a decent sized area that contains useful features for matching on. As long as you've got enough overlap to contain 10-20 good control points, I'd call it fine.
On colors, one guide I saw said to start shooting with the sun (or main light source) 90 degrees to one side and use that shot as the metering point. I set my camera to manual and pick out the settings first, then flip the focusing to manual, as well. So far, that seems to have worked out for me.
Jon |
|
|
|
06/04/2006 10:33:56 PM · #799 |
| Thanks for the pointers, Jon. It at least gives me some things to start my experimenting with. |
|
|
|
06/04/2006 11:52:06 PM · #800 |
Originally posted by Prism:
Thanks for the pointers, Jon. It at least gives me some things to start my experimenting with. |
Your panos look good. What you uploaded look like they have pretty good luminosity and color blending, but at the smaller sizes it is a little hard to tell.
The most important thing to remember from fracman's comments is that he had some images with poor overlap and also had issues with his merges. Those two things are related. Be sure to always have plenty of overlap.
My latest panorama odyssey... I decided to give Hugin a little tougher test to see how it would perform. I gave it a two row 14 image panorama taken near Sedona I had not previously put together.
I wanted to know if it could figure it out two rows on it's own without me doing anything special with control points. So first I tried stitching a 2 row 4 image pano from one end. Even after very close inspection the merge was perfect and Hugin figured out the control points by all by itself.
So I decided to put the whole thing together. It came up with 430+ control points for the merge. They looked fine to me so I optimized and decided to look at the preview. There were some funky things, the preview display was warped and the calculated height and width of the display was 360 degrees by 180 degrees which is not correct. It might be about 120 degrees wide or so, not more. Anyway I want to seee what it would come up with and let it go... After while I could see that it was responding very slow, so I double the amount of virtual memory...
4 hours later I decide to kill PTStitcher. When I did that it leaves all it's temp files and I could see that 11 of 14 temp files took about 1 minute to build. The last three had exponentially increasing amounts time. The last one was still being bult when I canceled the task but had been running over 3 hours itself.
There appears to be plenty of available free disk space but this PC has only 512 megs of RAM. I will do normal cleanup maintence, reboot and try again... only this time with only 10 images. :)
|
|
