MaskedStrechTransform

[José Manuel Pérez Redondo]

Hola

Estoy practicando con el script MST, después de leer el texto de Sonnenstein (31/12/2007) y tengo dudas sobre algunos valores.

1. No acabo de coger el truco al Target Median. Hago hasta 100 iteraciones con diferentes valores, pero no acaba de estirar la imagen. Siempre tengo que sacar más señal con curvas.

2. Hay alguna estrategia para decidir si utilizar wavelets o Convolution.

Gracias

[Andres.Pozo]

1. No acabo de coger el truco al Target Median. Hago hasta 100 iteraciones con diferentes valores, pero no acaba de estirar la imagen. Siempre tengo que sacar más señal con curvas.

Te recomiendo que antes de aplicar el MaskedStretch apliques un HistogramTransform lineal para recortar las sombras. Deberías mover el control de sombras hasta el punto en el que empieza haber valores procurando no perder ningún valor (o menos del 1%).

[dhalliday]

Any chance  of an English translation...? (Juan ?)
You Spanish guys sound pretty clever...don't want to miss anything !

Dave

[Andres.Pozo]

My answer in spanglish:
Before applying the linear MaskedStretch you should apply an HistogramTransform to clip the shadows trying to not lose more that 1% of the pixels.

[sleshin]

Dave,

You can translate any foreign language using this Google page: http://translate.google.com/#

In Windows, just right click on the web page you want to translate and click "save page as". I use default file name and file type and save it to desktop. Then open Google link and click on "upload a document" and Browse to the desktop file you just saved. Choose Spanish to English is this case and click on Translate. Then like magic you are instantly multilingual.

Steve

[José Manuel Pérez Redondo]

Hola

perdonad por no responder antes, problemas informáticos.

Andrés. Cuando aplico el script el histograma lo tengo totalmente pegado a la izquierda. Me refiero a que no lo he estirado. Es el primer paso que hago.

(google translation)

Hello

pardon me for not responding earlier computer problems.

Andrés. When I apply the script I have the histogram completely glued to the left. I mean, I have not stretched. It is the first step I do.

(and the first post)

Hello

I'm practicing with the script MST, after reading the text of Sonnenstein (31/12/2007) and I have doubts about some values.

1. I do not quite catch on to the Target Media. I make up to 100 iterations with different values, but not just stretch the image. I always have to get more signal curves.

2. There is some strategy to decide whether to use wavelets or Convolution.

Thanks

[José Manuel Pérez Redondo]

Bueno, sigo insistiendo a ver si alguien me ayuda

En esta imagen está el resultado de 3h40m de exposición de m42 y a su lado un mackstrech de 100 iteraciones. Encuentro que el resultado es muy pobre pero no veo mi error.

Os agradecería un ayuda

Gracias

José Manuel

Well, I keep insisting to see if someone helps me

This image is the result of exposure 3h40m of m42 and its neighbor a mackstrech 100 iterations. I find that the result is very poor but I see my mistake.

I would appreciate a help

Thanks

José Manuel

[georg.viehoever]

Hi,

I think you need to do a HistogramTransform first.

Georg

[Carlos Milovic]

El problema es "target median". Se supone que es el valor final que quieres que tenga el cielo. Debe ser mucho más alto, en torno a 0.1


The problem is the "target median" value. It should be close to 0.1, since it is the spected value for the median, after the MST. Since sky background usually dominates the images, the median is a pretty good aproximation.

[José Manuel Pérez Redondo]

Gracias a todos.

Carlos, con 0'1 funciona mucho mejor, pero luego tengo que estirar la imagen con el histograma. ¿Lo hago bien o todavía tengo que mejorar los parámetros del MST?

Thank you all.

Carlos, 0'1 works much better, but I have to stretch the image with the histogram. Am I doing it right or I still need to improve the parameters of the MST?

[Carlos Milovic]

0.1 es un valor que a mi me funciona bien, cuando hay mucho cielo. Al final, esto es cosa de gusto personal, y de los requerimientos de la imagen. Prueba modificando ese valor. Más grande que 0.3 es poco probable que ande bien.



0.1 is a value that wors good enough for me. This is true when there is a lot of background sky. At the end, this is just a matter of personal taste, and what your image needs. Try higher values. It would be very unlikelly that it goes above 0.1, but again, this is up to you.

[José Manuel Pérez Redondo]

Gracias Carlos

Al final he realizado pruebas a 0'1 y 0'12, con 200 iteraciones y he quedado muy satisfecho.

Thanks Carlos

At the end I did proves with 0'1 and 0'2, with 200 iterations and I am very happy with the result

José Manuel

[dhalliday]

Excuse my ignorance here...but what is this feature..?
When is it used/what does it do..??
First I have heard of it...
Ahhh the labyrinth that is Pixinsight...!

Dave

[Simon Hicks]

Hi Dave,

Its been in PI for...well at least hundreds of years (well a couple anyway). It takes an unstretched image, then does a whole series of small incremental stretches whilst protecting the star edges at each stage. The aim is to try and minimise the star bloating that you can get if you do a large histgram stretch all in one go.

At least that's what i think it does....but someone may be able to give more details.

I tried it hundreds of years ago and compared it with a single full stretch....and to be honest I couldn't tell any difference. Maybe I was doing something wrong. I like the concept, so I would be interested if others have actually found it makes a difference....i.e. they have done a direct comparison between using it and using a normal stretch.

Cheers
         Simon

[dhalliday]

Simon
Thanks..
Perhaps a volunteer can show us a "before and after" (with/without) pic to demonstrate further...?