Completed stitching all the panels I shot at the star party.
The thumbnail posted is only a crop of the area. I just can't fit it all in 200KB.
The 80% res of the full uncropped field is here [21.2MB - 5622x4931px]
I also made a smaller for download here [10MB - 3514x3082px]
and an even smaller version here [3.16MB - 1757x1541px]
Sorry the files are huge but it's a massive area with a pretty good image scale. Thanks for looking.
Last edited by multiweb; 18-05-2013 at 08:30 PM.
Reason: Updated files and thumbnail
Thanks guys. I hope you've managed to look at the big one. The stars and colours are a lot better in the big one. Smaller versions suffered clipping and contrast issues by reducing the file size. I think PS doesn't keep the original histogram when reducing. Maybe I should bin the FIT files to present smaller areas or save to TIFF 16bit and bin in CCD Stack preserving the flux. Will play around with it. Not happy with the luminosity and colours the reduced files came up with. Stars look clipped without colors and the black point shot up.
Very impressive Marc. Subtle processing too - you haven't gone overboard. The skill in knitting this together and presenting so well is not easily appreciated.
Very impressive Marc. Subtle processing too - you haven't gone overboard. The skill in knitting this together and presenting so well is not easily appreciated.
Thanks Rob. Stitching is one thing I'm used to now. It's second nature. But it's the first time I've had this issue with presenting a very large area at a smaller scale. I've just done a bin 3x3 test in 16bit environment and it's chalk and cheese with the smaller files I posted, so due for a repro.
Well the proof is in the pudding. I think this will be of interest to people resampling their work at lower res for fast web display.
First thumbnail is reduced 50% from the orginal by doing a resize in Photoshop. TIFF 16bit. You clearly see that the stars become white clipped and the plack point is raised as a result of the increased contrast. Not what you want.
Second thumbnail is the same shot but this time I did a bin2x2 in CCD Stack then did some pixel math and divided the file by 4. The histogram is not clipped and the stars retained their color. (maybe not obvious in the compressed JPEG but obvious in the original)
So repeating the resizing only compounds the issue which I usually do so every smaller picture gets clipped again and again and gets worse.
I'll go through the various versions and update the previous pics in the post.
As for your clipping, normal bicubic resizing shouldn't clip your whites if they weren't already clipped to begin with. Unsharp masking, however, will clip them - especially if your images were already high contrast to begin with. If you use the "Bicubic - best for reduction" mode, it behaves as if it applies USM to the result - which might explain your clipped whites? Are you applying any levels/curves/USM to your resized version?
As for your clipping, normal bicubic resizing shouldn't clip your whites if they weren't already clipped to begin with. Unsharp masking, however, will clip them - especially if your images were already high contrast to begin with. If you use the "Bicubic - best for reduction" mode, it behaves as if it applies USM to the result - which might explain your clipped whites? Are you applying any levels/curves/USM to your resized version?
Thanks Dave. The clipping occurs in PS when resizing down. Even if the pic you're resizing doesn't have any clipping to start with. Once it's clipped you cannot recover the clipped data via curves or level.
Thanks Dave. The clipping occurs in PS when resizing down. Even if the pic you're resizing doesn't have any clipping to start with. Once it's clipped you cannot recover the clipped data via curves or level.
Yep, what I'm getting at is that this clipping shouldn't normally be happening to such a noticeable extent - since it's clearly modifying the content of the images, not just resizing it.
I've attached histograms before vs after resizing using the various modes in Photoshop CS5. You can see that only "Bicubic sharper" results in the clipping spike that you're talking about. You'll get similar results with "Bicubic best for smooth gradients" mode followed by unsharp masking.
Yep, what I'm getting at is that this clipping shouldn't normally be happening to such a noticeable extent - since it's clearly modifying the content of the images, not just resizing it.
I've attached histograms before vs after resizing using the various modes in Photoshop CS5. You can see that only "Bicubic sharper" results in the clipping spike that you're talking about. You'll get similar results with "Bicubic best for smooth gradients" mode followed by unsharp masking.
Actually now you've mentioned it I did this in CS6. I'll try doing the same in CS5 and check the results.
PS: You're right. CS5 doesn't do it. CS6 does. I'll have to google that up see what's causing it.
Maizing Stuff Marc, I cant get over how you do such a veriety of stuff, and putting panels together as this ! Very inspiring for some but over this ducks head mate,
champion stuff &
Maizing Stuff Marc, I cant get over how you do such a veriety of stuff, and putting panels together as this ! Very inspiring for some but over this ducks head mate,
champion stuff &
Thanks Bob. It's actually very easy to stitch stuff. Not much skills involved. You can use registar to align the panels then use PS to blend them in. It's pretty much all automated. Just little touch ups there and there. The trick is to get the right color balance in the different panels as close as possible before joining them. But you do that by the numbers too so no guess work. Just practice.
Yep, what I'm getting at is that this clipping shouldn't normally be happening to such a noticeable extent - since it's clearly modifying the content of the images, not just resizing it.
I've attached histograms before vs after resizing using the various modes in Photoshop CS5. You can see that only "Bicubic sharper" results in the clipping spike that you're talking about. You'll get similar results with "Bicubic best for smooth gradients" mode followed by unsharp masking.
Figured out what it is after posting on the adobe forum.
In CS6 the bottom drop down defaults to bicubic (automatic) in the image resize window. I never noticed it before TBH. Changing it to Bicubic (best for smooth gradients) sorted out the issue although it defaults back to its original automatic settings. No clipping now. In CS5 it defaults to Bicubic (best for smooth gradients) which is why it worked in one but not the other.
Yeah, I had a rummage around your sky drive images and it seems to be a characteristic of your work. Your processing is very subtle, and it works.
I couldn't even begin to think about how to achieve what you've done here.
Truly beautiful.
Really bummed I never got the star party this year. Did you ever put together the time lapse images that you took last year? I'd love to see them. And did you ever get a pair of those bino's from bintel?
