Author Topic: Algunas preguntas sobre ColorCalibracion/ Some questions about ColorCalibration  (Read 3074 times)

Offline ManoloL

  • PixInsight Addict
  • ***
  • Posts: 220
    • View Profile
Hola a todos:

El pasado fin de semana aprovechando la ausencia de luna y nubes estuve fotografiando a M31, con mi nueva cámara QHY8.
Fueron exposiciones de 30s, 1m, 2m, 5m y 15 minutos. De estas ultimas que son la base de la imagen obtenida obtuve 9 exposiciones entre las dos noches.
Las imágenes las calibré con darks, flats y bias y las apilé con DSS, usando el superpixel para la debayerización.
El resto de las operaciones las realice con la última versión de PixInsight (1.5.6), llegando a esta imagen.

http://fotografiaastronomica.com/files/posted_images/user_41_hdrandrv1j.jpg

Aunque la imagen es una de las mejores que he logrado no estoy satisfecho con ella, fundamentalmente en lo que se refiere a los colores.
Las imágenes clásicas de M31 presentan la periferia de la galaxia de un tono azulado que no he sido capaz de conseguir.

Por ello estoy reprocesando la imagen tomándome el tiempo que sea necesario e intentando no realizar pasos en falso en el procesado.

Al llegar a ColorCalibración me han surgido una serie de dudas:

He llegado a una imagen después de aplicar el script HDR para ampliar el rango dinámico, partiendo de las imagenes apiladas con DSS que he registrado entre si con StarAlignment. Esta imagen tiene unos gradientes evidentes al aplicar STF

¿Hay que aplicar el DBE como primer paso antes que ColorCalibración?

¿Hay que aplicar BackgroundNeutralización previamente a ColorCalibración?

Esto parece deducirse de lo escrito en http://pixinsight.com/forum/index.php?topic=1105.0 exactamente cuando Juan escribe:
Después de BackgroundNeutralization (que es un paso previo obligatorio para poder alcanzar precisión en la calibración), la copia de ......

¿Efectúa el propio ColorCalibración el BackgroundNeutralización como parte de su proceso y no es necesario realizarlo previamente?

A continuación intentaré, con el uso de un traductor automático y mi intuición corrigiendo las incoherencias que detecte, traducir al inglés el texto:

   
Hello everyone:

Last weekend, taking advantage of the absence of clouds and moon, I photographed at M31 with my new camera QHY8.
Exposures were 30s, 1m, 2m, 5m and 15 minutes. Of those (15 m) that are the basis of the image obtained 9 exposures between the two nights.
The calibration of the images made with the darks, flats and bias and stacked with DSS, using the superpixel for debayerización.
The remaining operations are carried out with the latest version of PixInsight (1.5.6), to this image.

http://fotografiaastronomica.com/files/posted_images/user_41_hdrandrv1j.jpg

Although the image is one of the best I've achieved I'm not happy with it, primarily in regard to color.
The classic images of M31 show the periphery of the galaxy in a bluish tone that I have not been able to achieve.

I therefore reprocessed the image using the necessary time and trying not to make false steps in processing.

Upon reaching ColorCalibración I have emerged a number of questions:

I have an image after applying the script HDR to extend the dynamic range, using images stacked with DSS and aligned with StarAlignment. This image has a gradient that appears to apply STF.

Should we implement the DBE as a first step before ColorCalibración?

Should we implement BackgroundNeutralización before ColorCalibración?

This appears in the written http://pixinsight.com/forum/index.php?topic=1105.0 exactly when John writes:
[color = green] After BackgroundNeutralization (which is a mandatory initial step for the sake of calibration accuracy), the above......[/ color]

Does the ColorCalibración the BackgroundNeutralización as part of its process and is not necessary previously?

Best regars

      

 Editado: Juan, perdona por trasformarte en John. Como puedes suponer ha sido el traductor que me lo ha "colocado" y no me he dado cuenta hasta después de publicarlo.
« Last Edit: 2009 August 01 01:24:52 by ManoloL »
Saludos.

Manolo L.

Offline Simon Hicks

  • PixInsight Old Hand
  • ****
  • Posts: 333
    • View Profile
Hi Manolo,

I am pretty sure you should apply the DBE before the colour calibration or background neutralisation. DBE will usually change the colour of the image anyway because it calculates different background functions for each colour. If you apply colour correction / background neutralisation you will get "corrected" results....and then if you apply DBE afterwards then the correction is lost.

I think another way of looking at it is that you first want to get rid of all the artifacts in the image before getting the colour right. So DBE gets rid of the gradients, and then you apply the neutralisation followed by colour correction. Think of DBE more like an extension of all the corrections made in DSS right at the beginning of the process.

Cheers
         Simon

Online Juan Conejero

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 6715
    • View Profile
    • http://pixinsight.com/
Hi Manolo,

I'll answer your post in English, as I think you'll understand it without problems, and mainly because you've posed some questions that can be of interest to everybody. Let me know if you want me to translate this into Spanish. No problem at all with calling me John, by the way; just don't get accustomed to do that (joking) :)

The image is very nice, and your HDR composition work is also good (the core could be improved a little bit, though), but, in my opinion, color balance is clearly wrong.

In first place, direct answers to your questions:

Quote
Should we implement the DBE as a first step before ColorCalibration?

In general, yes, as Simon has pointed out. Gradients pose a twofold problem to implement a valid color calibration. On one hand, gradients make it impossible to evaluate a mean background for the whole image, which is necessary as a reference value for the color calibration routines. In addition, since gradients also change the midtones and the highlights irregularly, they can prevent color calibration routines from computing consistent white balance values for the whole image.

Note that the same applies to image calibration: if the image has not been flat-fielded very accurately, color calibration routines won't work accurately either, for the same reasons explained above.

Quote
Should we implement BackgroundNeutralization before ColorCalibration?

Yes, always. This is true even after DBE correction (which tends to yield a neutral background after subtracting a valid background model). BackgroundNeutralization is a fast routine that can safely be applied with default parameters. It guarantees a truly neutral background in the statistical sense that ColorCalibration requires.

Quote
Does ColorCalibration the background neutralization as part of its process, so it is not necessary as a previous step?

Not at all. ColorCalibration just works with the existing background. Color calibration and background neutralization/equalization are different, not necessarily dependent tasks (you know, this is part of PixInsight's philosophy: divide and conquer).


Now, regarding your M31 image, I'd bet the problem is in the white reference you've used. How did you apply ColorCalibration?

If I were you, I'd define three or four previews covering the whole galaxy (located diagonally over M31). I'd define two additional previews, one at each side of M31, over background regions. Then I'd use David's PreviewAggregator to gather the previews as two single images, which would be the white and background references, respectively. I think if you apply ColorCalibration with these references the result should be basically perfect; this is in fact a pretty easy target for color calibration, since the references are easily separable.
Juan Conejero
PixInsight Development Team
http://pixinsight.com/

Offline ManoloL

  • PixInsight Addict
  • ***
  • Posts: 220
    • View Profile
Hola a todos:
Gracias por las respuestas.

Juan:
No tengo ningún problema en leer textos en inglés.
Tengo muchas dificultades para escribir en inglés.
Por ello prefiero hacer las preguntas en español, aun intentando añadir una traducción al ingles que nunca estoy seguro si es correcta.
Antes de recibir la clarificadora respuesta de Juan ya había realizado una nueva versión de la imagen en la que en gran parte he seguido el procedimiento que indica, aunque no completamente:
No he realizado el backgroundNeutralización despues del DBE, pensando erroneamente que lo hacia el propio  ColorCalibración, dado que hay que proporcionarle muestras del fondo de cielo.
Como muestra del blanco he usado toda la imagen.
La muestra de fondo si la he generado con el Agregator de David.
El resultado ha sido este:

http://fotografiaastronomica.com/files/posted_images/user_41_hdrandr70_dbectrdacl.jpg

Repetiré el proceso siguiendo la secuencia indicada y  obtendré nuevos resultados.

Por cierto en estos procesados tuve que modificar los valores por defecto del script HDRComposición pues me generaba un anillo azul alrededor del nucleo de la galaxia.

http://fotografiaastronomica.com/files/posted_images/user_41_andr15m9e2df2f.jpg

Experimentando vi que desaparecía al bajar de 80 a 70 el Highlight limit (high) y el Binarizing treshold de la Composición Mask.
No termino de estar satisfecho con los resultados, aunque han mejorado y agradecería información sobre los parámetros del script

Editado: He encontrado la información de los parametros del script en una nota de Vicent Peris en este hilo:
http://pixinsight.com/forum/index.php?topic=1074.0;prev_next=next#new

Saludos.



Hello everyone:
Thanks for the replies.

Juan:
I have no problem in reading texts in English.
I have great difficulty writing in English

Therefore I prefer to ask the questions in Spanish, while attempting to add an English translation that I'm never sure if it is correct.
Before receiving the illuminating response of Juan, I had already made a new version of the image in which I followed largely the procedure indicated that, although not completely:
I have not done backgroundNeutralización after DBE, thinking erroneously that ColorCalibración did it, given the need to provide samples of the background sky.
As sample in the white I used the entire image.
The sample of the background if I've done it with David's Agregator
The result was this:

http://fotografiaastronomica.com/files/posted_images/user_41_hdrandr70_dbectrdacl.jpg

I repeat the process following the sequence indicated, and get new results.

Certainly in these processes I had to modify the default script HDRComposición because it created a blue ring around the nucleus of the galaxy.

http://fotografiaastronomica.com/files/posted_images/user_41_andr15m9e2df2f.jpg

Experimenting, I saw that the ring disappeared when changing from 80 to 70 the Highlight limit (high) and the Binarizing treshold of the Composición Mask.

I am not satisfied with the results, although they have improved and I would appreciate information on the parameters of the script.

Edited: I found the information on the parameters of the script in a note by Vicent Peris in this thread:
http://pixinsight.com/forum/index.php?topic=1074.0;prev_next=next#new

Regards
« Last Edit: 2009 August 02 12:30:54 by ManoloL »
Saludos.

Manolo L.

Offline ManoloL

  • PixInsight Addict
  • ***
  • Posts: 220
    • View Profile
Hola a todos:

Me es grato comunicaros que mi M31 ha sido publicada como la imagen del mes en la revista Astronomia.
No me cabe ninguna duda que eso no hubiera sido posible sin PixInsight, pese a mis deficiencias en su manejo.

Saludos.

Hello everyone,

I am pleased to inform you that my M31 has been published as the image of the month in the journal Astronomia (Spain).
I have no doubt that this would not have been possible without PixInsight, despite my shortcomings in its management.

Greetings.
Saludos.

Manolo L.

Offline Nocturnal

  • PixInsight Jedi Council Member
  • *******
  • Posts: 2727
    • View Profile
    • http://www.carpephoton.com
Congratulations! Always nice when you get published. I'm still waiting for the first time :)
Best,

    Sander
---
Edge HD 1100
QHY-8 for imaging, IMG0H mono for guiding, video cameras for occulations
ASI224, QHY5L-IIc
HyperStar3
WO-M110ED+FR-III/TRF-2008
Takahashi EM-400
PIxInsight, DeepSkyStacker, PHD, Nebulosity