Wierd STF Permanancy

[rdryfoos]

Hello,

When I apply the STF, the stretch to the image is supposed to be non permanent.  However, when I apply the STF to the histogram and then render the stretch permanent by transferring the histogram triangle to the image, the stretch is different depending whether I used the STF with the channels linked or unlinked.  If the STF is not permanent, but just gives a visual indication of what ab stretch would be like, then why does uing the channel link button on the STF produce a different result when applied to the histogram

Then...which version is the right one--transferring the STF to the histogram with channels linked or inlinked.

Rodd

[pfile]

STF has 2 modes, in one mode it locks all 3 channels together, computing one common stretch which is shared between all channels. which channel controls the stretch is dependent on the image statistics.

in the other mode it computes the stretch individually for each channel.

not sure what any of this has to do with permanence. you are just seeing the results of two different methods for computing the stretch.

if the linear image was properly color balanced, then the only correct method is channels linked. otherwise what STF is essentially doing is computing a color balance based solely on image statistics, which is very likely to look OK, but be wrong vs. how PCC or ColorCalibration computes the color balance.

if you are just messing around or do not care about color balance, then applying the unlinked STF to the HC and then to the image might be OK for you.


rob

[msmythers]

Rodd

If you have not color balanced your image then the linked channels STF can be very different from the unlinked. I've attached the section of the STF documentation that deals with linked and unlinked channels. Which is the correct use is up to you and how you are processing your image. Unlinked STF does not do a 'true' color balance of an image like the combination of BN/CC or the new PCC tool. That being said if your ok with how your image looks with unlinked STF then it's available to be used.



Mike

[rdryfoos]

I did a Photometric calibration and the linked is way different than the unlinked.  The photometric calibration worked--according to PI.  And the weird thing is the unlinked version gave a histogram where all channels were aligned when I transferred it to the histogram.  The linked channels were way off--the image was all red.

[msmythers]

Rodd

Did you reset STF after applying your stretch?


Mike

[rdryfoos]

Rodd

Did you reset STF after applying your stretch?


Mike

Not sure what you mean.  I did my DBE which worked well, tyhen the PCC, which seemed to work well.  But the STF looked very different blinked or unlined.  Linked it looked all red and unlined looked more correct.  And the histogram looked very different after I made the stretch permanent--I treid both. 

[rdryfoos]

I know kit doesn't mean much without both--but here is the finished image done with the unlinked SGTF transfer.  It has been processed since that stretch, of course.  But teh color separationsn you see were pretty much defined.  When the channels were linked--only red was visible. 

https://www.astrobin.com/322446/M/?nc=user

[pfile]

I did a Photometric calibration and the linked is way different than the unlinked.  The photometric calibration worked--according to PI.  And the weird thing is the unlinked version gave a histogram where all channels were aligned when I transferred it to the histogram.  The linked channels were way off--the image was all red.

that is what an unlinked STF does - blindly aligns the channels.

i would say that your PCC did not actually work if the image had a red cast. you might try ColorCalibration as a sanity check...

rob

[msmythers]

Rodd here is a very simple flow of STF, PCC and the final stretch.

Apply STF to image linked, very grren in this case. Apply PCC, reapply STF linked. Image looks correct. Apply that STF to Histogram and stretch. You now reset STF(lower right of STF) and you have the normal image stretched.
Notice how STR changes with every step. STF needs to be applied each time and once stretched reset.



Mike

[rdryfoos]

Rodd here is a very simple flow of STF, PCC and the final stretch.

Apply STF to image linked, very grren in this case. Apply PCC, reapply STF linked. Image looks correct. Apply that STF to Histogram and stretch. You now reset STF(lower right of STF) and you have the normal image stretched.
Notice how STR changes with every step. STF needs to be applied each time and once stretched reset.



Mike

OK--yes I see--that is how I do it.  But normally for me, if CC does not work, then the linked SFT produces a wildly different outcome--it never results in aligned channels.  In this case, the unlinked SFT produced absolutely correct results (I processed this image about 50 times as usual) and in previous tries, the linked SFT looked like the unlinked in this one.

[msmythers]

Ok Rodd

I would definitely try what rob said. That should work.

While PCC is very good, some images can give it grief when using the defaults. Minor changes in the Photometry parameters can make big changes. Just for grins you might try changing the automatic settings. I have had some success(and failure) by changing the Limit magnitude to 10 and the Aperture to 10. Can't hurt. You can also try using PCC on a crop of your image and see if that changes anything. There are a lot of variables to deal with.


Mike

[rdryfoos]

Ok Rodd

I would definitely try what rob said. That should work.

While PCC is very good, some images can give it grief when using the defaults. Minor changes in the Photometry parameters can make big changes. Just for grins you might try changing the automatic settings. I have had some success(and failure) by changing the Limit magnitude to 10 and the Aperture to 10. Can't hurt. You can also try using PCC on a crop of your image and see if that changes anything. There are a lot of variables to deal with.


Mike

I reprocessed it using BN and CC--came out reddish that way too.  The problem is I like it with the blue differentiated a bit more. 

[pfile]

when you run CC do you have the "output white reference mask" and "output black reference mask" turned on? that will give you an idea if in fact the way you have the thresholds set pick up any pixels (or pick up too many pixels...)

rob

[Alejandro Tombolini]

Hi Rodd, maybe this or this video may help.

Saludos, Alejandro.