Experiences with PixInsight LE

[Carlos Milovic]

Hi all

Please tell us here your experiences with PixInsight's interfase and tools. Was hard to learn how to use them? Do you like the documentation? Tell us , as well, any suggestion about both topics (program & documentation). Your comments are highly apreciated! =)

[sfernandez]

Primero les comento que he trabajado con photoshop durante muchos años, generalmente tratando texturas e imágenes fotográficas (no astrofotografía). Soy en extremo principiante en este hobby y estoy ahorrando para mi primer equipo (tele principalmente, puedo pedir prestada una Canon eos de mi hermana cuando llegue el momento)

Por lo tanto estuve probando PixInsight con fotografías nocturnas hechas con una Sony dsc v-1 y con imágenes del dss (ésta semana empecé con eso).

La interfaz no me pareció compleja, pero al tener tantas opciones, es un poco abrumadora al principio. De todas formas, luego de unas semanas se torna natural su uso, y al volver al photoshop, las ventajas son obvias, como ser los previews modes y process icons, que ahorran muchísimo tiempo.

Los histogramas y curvas son en extremo completos. Imágenes con las que sufrí muchísimo en ps son fáciles de tratar en PI, especialmente en la corrección de colores.

Los filtros de ruido son sencillos de usar, especialmente scnr y sgbnr. A modo de comentario, a las fotos nocturnas obtenidas con la sony le aplico scnr average green y average blue y desaparece todo el ruido "grueso" y los "halos" alrededor de fuentes de luz, fabuloso.

Todavía me faltan explorar muchas opciones más, que recién estoy descubriendo al tratar de procesar una "astrofoto" real.

Sebastián

pd: Perdón Carlos, recién llego de trabajar y no tenía ganas de escribir en inglés.

[Carlos Milovic]

Da lo mismo si escribes en español o ingles. Lo que nos importa es recibir el "feedback" de los usuarios. Muchas gracias por entregarnos tu opinión.

[Jeff Randall]

The initial GUI experience is quite easy, clean and intuitative.  But after printing out and skimming the approximately 336 pages of documentation, there is a LOT more to the program than meets the eye [as my naive suggestions posted earlier in this forum prove]!!

Overall the documentation is clear and well written with good and simple explanations although it sometimes drifts into serious techno-babble -- at least to me.  I am unfamiliar with the jargon of the astrophoto community so maybe I am way off base with this observation and that most of your customers is familiar with the terminology.

Here is one example that illustrates how using more familiar terms makes the program more accessable to wider range of potential users.

Process Icons and Process Containers:  These elements were initially difficult to comprehend, but once I mentally "translated" the jargon into a more familiar frame of reference I did much better.  The fog lifted in my mind once I saw Process Icons as Windows short-cuts to processes and that the processes have sticky settings and saw Process Containers as equivalent to PhotoShop actions (or other program's macros or batch scripts).

Hope this gives you some ideas.

Jeff Randall
Preston, Washington
USA

[Zie]

I have to admit, I am not intricately aquainted with noise reduction techniques, so the learning curve has been pretty tough. However, I absolutely love the results I get with it. Thanks for making such a great program!

[macca]

G'day,

I've only started using PixInsight on a few of my images and whilst there is a steep learning curve I think you're onto a winner. There are a lot of difficult concepts to grasp, and it certainly helps being an engineer to understand them!

There are a few things that could be improved (such as a workflow and a few GUI optimisations) but they will come with time as the product matures. But the most important thing I think you can further improve is the documentation. To me, this is the key to helping new users overcome the initial hurdles. Overall the documentation is pretty good, but sometimes it tries to explain things too technically and not always from a users point of view (ie what effect will changing this value have on my picture?). The extra information is useful, but it should be left towards the end of a section. I should only have to read the first few lines of a paragraph to get a rough understanding of whether I need to change a particular option to get the effect I am after.

As a small example consider the short paragraph under the heading "Smoothing Filter Kernel Size" found here here. This could be improved by first stating that the value of this number is related to the noise scales you are trying to reduce. ie, a smaller number works better on high frequency noise. It's only a small thing but with so much documentation to sink in, it means the difference between skimming the sections that are not relevent and having to read absolutely everything.

On the positive side, I think the mouseover images in the documentation makes a real difference. I don't think there can be enough of them to help illustrate a point.

Great work so far guys. I'm really looking forward to PixInsight Standard whenever it's ready.

Craig.

[Jack Harvey]

Have only recently been using Pixinsight and am impressed with the power of this program.  I feel the Histogram and Curves are much more exact and sensitve than PS.  I do like the real time Preview.  On my calibrated moniter the colors and saturations are different between the RT Preview and the Image, but its helpful to get an idea of what results to expect when you execute on the image.  Am still learning wavelets and using the DBE and the noise reduction tools.  Color saturation is an area that needs to be improved some in my opinion.

I have put together a couple of images using CCDStack and Pixinsight only and the results are very encouraging.  Am very interested in the standard edition.

Jack Harvey
www.starshadows.com

[Juan Conejero]

Hello Craig,

Thank you very much for your warm words.

I understand what you say about our documentation, and I completely agree. In many cases we have not had the necessary perspective to be able to put ourselves in the user's place.

Writing a comprehensive and useful documentation for something like PixInsight is not the easiest task, as I'm sure you know very well. Now we have accumulated a lot of experience, especially feedback from users, that hopefully will let us produce a much better documentation for the next versions.

I appreciate the example you've excerpted from the wavelets section; very helpful to catch your idea.

Regarding the GUI, wait to see our next implementations. We are rewriting the whole application from its foundations, and the entire user interface is now based on the Qt library by Trolltech (www.trolltech.com/qt/). I'm sure you'll love it as much as we love Qt and its awesome power. As you probably suspect now, yes, we plan to port the application to Linux and Mac OSX platforms, including 64-bit platforms. The initial release will be for 32-bit Windows, though.

You probably will be interested to know that PixInsight Standard will be an open system. All processes, file formats and processing interfaces will be implemented as installable modules. There will be a language-independent, low-level API, and a high-level API, that will be a C++ class library. The high-level API will be freely available.

Again thank you for your interest. I'd like to know more on your experience with PixInsight, so please feel free to comment on here.

Best,

Juan

[Juan Conejero]

Hi Jack,

Thank you. I am glad to know that you like our Curves and Histograms implementations. We always have thought that both processes are extremely important, and in many cases they are not implemented with the deserved care and accuracy.

Regarding the real-time preview, take into account that PixInsight is a native 32-bit floating point application. This means that the internal working dynamical range is huge and cannot be represented by any output device, especially on a monitor. The Real-Time preview window uses an internal 8-bit representation based on look-up tables (LUT) to speed up screen rendering. When you preview an aggressive histogram or curves transformation, especially a saturation transform, it is simply impossible for a LUT-based system to generate an accurate rendition, so it is normal to see some differences with the actual processing result.

Our next implementations will provide an optional 16-bit real-time preview, which will be much more accurate, although slightly slower.

The standard version includes a specific color saturation process that you'll love. However on the Curves window you have a special saturation curve that can work very well.

We are working extremely hard on the standard version. As I've said in a previous post, we are rewriting it from scratch, and the volume and complexity of new functionality are overwhelming. We expect to have a first version ready for public release on fall - early winter. Perhaps we might have a preview release before summer, but I don't dare to ensure that.

Please let us know when you have some of these images ready to be shown. By the way, you guys have a wonderful setup there at SSRO... What a dream machine that remote 16" RC! I like your alien mascot as well

Cheers,

Juan

[Jack Harvey]

Juan  I discovered the Saturation Curve tool when re doing an old M 27 and found it works pretty well.  I understand the complexities of the preview window now.  As i did say it is an indicaiton of how you are adjusting the image and with experieince you can correlate the two pretty well.  Please keep us posted of the programs progress.  Jack

[Carlos Milovic]

Hi Jack

what you said is true: RT Preview is only an approximation to the real results. We encourage you to try Histograms and Curves adjustements on a common preview before applying it to the image (as said in the documentation).

Thanks for the feedback!

[Jack Harvey]

Yes I have been trying that, but I also find with some experience you can get a rough idea of how the image will look from the RT preview with experience.  For instance you refrain from making that last histogram improvement that the RT preview window would suggest you should do and then the image is real close.  Jack

[Carlos Milovic]

You are right. Most of times the RT Preview does an excelent job. But, many times, when facing aggressive changes, it fails completely. The best example is a high midtones balance value (well, or very low...), as we have to apply on RAW CCD images.