COPYRIGHT, PLEASE NOTE

All the material on this website is copyrighted to J-P Metsavainio, if not otherwise stated. Any content on this website may not be reproduced without the author’s permission.

Have a visit in my portfolio

Vieraile portfoliossani

Monday, November 9, 2009

The power of Tone Mapping

I like to show an example of Tone Mapping procedure.
I have used this method now about two years with very good results.
A PDF-document of the Tone Mapping can be found HERE.
NOTE! Images have to be perfectly calibrated and gradient free to this method to work.
-
The image of Sharpless catalog objects Sh2-223, 224 and 225 shows two extremely dim supernova remnants in a same Three degrees field of view.
Original Post about this image can be found here:
http://astroanarchy.blogspot.com/2009/04/sh2-223-224-and-225-lightened-up.html
As can be seen, in Stacked and Stretched 16h H-alpha image, nearly nothing can be seen under a massive Star field.


In the next image objects can be clearly seen after a Star removal procedure.
If Stars are removed with care, no information from actual target is lost.
If there is some minor details lost during a procedure, they are placed back automaticly, when Stars are placed back to a image.


When Stars are removed, it's easy to use Levels and Curves to boost actual data. There is now no need to worry about bloating Stars.
Since Stars has no relevant color information in a Narrowband image, this method can be used. In a Broad band RGB-image stars has real colors.
The same method can be used for RGB-image as well, if Star information is handeld correctly and replaced back to an image after manipulation.

Stars are placed back as a "Screen mode" under the PhotoShop


A "Color Map", from S-II, H-a, and O-III channels (Starless)



This image has all the color information from the Starless channels combined to a HST-palette.
I call it to a "Color Map".
In HST-color palette, the idea is balance colors so, that no line is dominatring the color scheme.
Usually the H-a is the strongest channel and HST images tend to have a strong Green cast doe that.
With this method, channels are easy to stretch equal, since there is no stars to worry about.
Very weak signal can be used, but it's allways better to have enough signal to work with.


A final image

The final image combined from a Tone Map and Color Map images.
As can be seen in a image, there is no Purple halos around stars, as typically seen in a HST-palette images, doe the strong stretching usually needed for a weaker channels.


In this image Stars are placed back. Whan stars are in separate layer over the Starless image, the tranparency can be tweaked carefully to balance visibility of Stars and Objects.
There is several methods to combine Stars and Objects. Stars can be placed as a Lighten mode in PS or by using the Screen mode. The right method depends of imagetype and some testing is needed by a user. Images can be combined by simply stacking them to gether as well.
We can't use H-a as a luminance alone, since it doesn't containe any information from other channels.
the weak signal from S-II and O-III channels is added to a H-a luminance
by a method descripted in a Tone Mapping PDF-document.
The result can be seen in the image above.
-


A Tonemapping Procedure diagram.
Download the PDF-document of Tonemapping from HERE.



An Animation of the Tonemapping procedure


-
Please, if you have tested this method, send a comment here.
Direct link to a you Tone Mapped image could be nice as well.
I need some feedback to be able to further improve this Work flow.



12 comments:

Vesa Kankare said...

I've just started to use this method after learning how to do it. And I have to say it works fantastic! My best success so far can be found here:

[NGC7000]

Thanks for sharing the method!

J-P Metsävainio said...

Good to hear Vesa.
The image of NGC 7000 is stunning!
I like the delecate color scheme.

Jan Inge Berentsen said...

Hi J-P! I'm a great fan of your work and have been using tone-map since october. I am really fresh at this hobby and have a lot to learn but I have had enough clear weather to tone-map 3 images since October. The last ones are here:
IC1795 in HST and IC1795 in CFHT

What really puzzles me is that you manage to get very little noise from a relatively short total exposure time. Take your latest tabpole-image for instance. Wonderful, - with only 1800 sec total SII and 2400 sec OIII. How?

Also, everytime I add the enhanced Ha I find that the image gets really washed out and I have to correct with curves and saturation. Am I doing things right here?

Thanks for putting up the .pdf!

Regards,
Ginge

J-P Metsävainio said...

Hi Jan!

Your images looks great, very good work.
Low noise is combination of many things. KAF 8300 CCD has a low dark current, carefull calibration with good bisas subracted flats and a great number of subframes for bias and darks. Usually I shot about 100 bias frames, 20-30 darks and 20 flats. Only falts are reshoot every imaging session for every filter used.

In photoshop, I use NoiseNinja for noisereduction to Starless channels. In final image a mild noisereduction is added at the end.
NoiseNinja is a good commercial noise reduction software. If you going to use any noisereduction method, take it easy. Too much reduction and you image will look very unnatural.

Hope this helps,
J-P

Jan Inge Berentsen said...

Thanks for the tip J-P! I didn't know I was supposed to bias-subtract my flats. Do you also subtract biasfrom your darks even when you use darks that have the same sub-exposure time as you lights?

Ginge

J-P Metsävainio said...

Yes, flats need to be bisas subracted, or if long exposure time flats, then dark subracted with darks taken with same exp. time.

Dark frame includes bias information. If you subract bias from dark, you'll end up adding it back to images with flats!!

Bias is used with darks only to scale them, if shorter exposuretime is used for lights, than darks.

J-P

J-P Metsävainio said...

I like to add one thing here.
The "Tone Mapping" method works well only to images with perfect calibration!
It's powerfull method to dig out very faint information. If lights are not well calibrated, it'll dig out all the flaws too.

Jan Inge Berentsen said...

Great! That was some very important information 8-). Thanks again!

Ginge

Anonymous said...

HI yes i found this too be a useful strategy as well

e.g

http://www.pbase.com/njh542/image/127518938

best wishes

Nik

J-P Metsävainio said...

Thanks Nik,

NGC6888 looks really good!

J-P

nik said...

again just like to say what a powerful and versatile technique this

Ive just processed my latest image the Pelican Nebula and the final image was definitely much better than i could manage with more "conventional" processing techniques

im thinking of trying this with LRGB images too but not sure how to handle star colours yet


Pelican

J-P Metsävainio said...

Hi Nik,

Thanks for the feedback! What a beautiful and smooth Pelican, great work.

BTW. I did some reorganizations to a text of the article. There was some miss placed columns in it, sorry.

J-P