NGC Thumbnails

Sat Apr 15 03:27:03 UTC 2017

Doesn't seem to be a lot of feedback right now.  I guess it could be because of the weekend?  Would you like to post a few of the bad examples and their NGC or IC numbers and I will see if I can do anything better with them?

________________________________
From: Kstars-devel <kstars-devel-bounces at kde.org> on behalf of Jasem Mutlaq <mutlaqja at ikarustech.com>
Sent: Friday, April 14, 2017 10:34 AM
To: KStars Development Mailing List
Subject: Fwd: NGC Thumbnails

Folks,

We've been having discussions over generating NGC/IC thumbnails to be used in KStars detail dialog + observation planner. John, from INDI forums (http://indilib.org/forum/wish-list/2136-ngc-objects-thumbnails-for-object-planner.html?start=24#16162), spent a lot of time generating image file (200x200) for NGC and IC objects in KStars database. However, when inspecting the images, there were quite a few issues. Below is an email from John that discusses the limitation of the approach. I'd like to invite all the developers to chime in on their ideas on this topic.

---------- Forwarded message ----------
From: John S. <myxlflik at hotmail.com<mailto:myxlflik at hotmail.com>>
Date: Fri, Apr 14, 2017 at 2:50 PM
Subject: Re: NGC Thumbnails
To: Jasem Mutlaq <mutlaqja at ikarustech.com<mailto:mutlaqja at ikarustech.com>>

Jasem,

FYI - I have completed the IC pulls.  They can be gotten here.  https://1drv.ms/u/s!AnSnRFrNRtfXviLmrzlp6Ru6N7KZ

[https://r1.res.office365.com/owa/prem/images/dc-generic_40.png]<https://1drv.ms/u/s!AnSnRFrNRtfXviLmrzlp6Ru6N7KZ>
IC-Thumbnails.tar.gz<https://1drv.ms/u/s!AnSnRFrNRtfXviLmrzlp6Ru6N7KZ>
Shared via OneDrive

The images are straight from the NASA data.  The ones that are out of 'out of focus' are likely the objects that are very small (less than two arc minutes in size).  There is an automated way to pull them and compile them in RGB.  The problem is that you either have to use all three channels from the lower resolution DSS survey or use 2 of the DSS2 channels (Red and Blue) and use the Green from DSS.  It's relatively easy to pull the objects, but it does take time.

I made some choices on parameters when doing the pulls.

200x200 pixels was the minimum thumbnail size.  1000x1000 was the maximum.

2 arcminx2arcmin is the closest resolution.  This means all objects 2 arcmin or smaller will be a 200x200 thumbnail scaled at 100 pixels/arcmin

    I chose to scale this up linearly to 1000x1000 where a 10 arcmin object would be 1000x1000 and still be 100 pixels/arcmin

    For objects larger than this, they will be 1000x1000 where the scale will be the number of arcmin wide or tall the object is.

As for the quality, I know a couple of the images are just bad because of the survey data.  I know there is room for improvement here, but we have 13000 object thumbnails that we didn't before.  [😊]

As for how I did the pulls, I used excel to create a custom pull for each object using  the syntax needed by the skyview.jar file from the NASA SkyView virtual observatory site.  Please feel free to experiment with this yourself and see if you can come up with better results.  I will definitely look into this further for you, but it will likely be after this weekend.

Here is an example pull for a black and white image:

java -jar skyview.jar position='NGC 7000' survey=dss2r scaling=log smooth=7 pixels=1000 Size=1.66666666666667 quicklook=png nofits output=thumb-ngc7000

This pulls NGC 7000 from the DSS2R survey at 1000x1000 pixels and 1.66666 degrees (100 arcmin) of scale per side and downloads it as a png file.

Here is an example pull for compiling RGB images:

java -jar skyview.jar position='NGC 2024' survey=dss2r,dss,dss2b rgb scaling=log smooth=7 pixels=1000 Size=0.5 quicklook=png nofits output=thumb-ngc2024

This pulls NGC 2024 using DSS2R for RED, DSS for GRN and DSS2B for BLU and combines them in RGB at 1000x1000 pixels and a scale of 30 arcmin per side.  This will download as a jpg and will have to be converted to png manually.

Try out a few pulls and see what you think.  The skyview.jar file has to be in the directory you run the commands from.  I have no doubt we can do more work on this.  I suspect some of the objects will do color just fine, but I know I ran into significant color issues with some objects.  NGC 7000 we particularly nasty.  I suspect data from some of the other surveys was missing and the channels just didn't come out properly.  It's possible that some of the other surveys have the data, but I did not look into it yet.

Let me know what you think and how you would like to proceed.

John

________________________________
From: Jasem Mutlaq <mutlaqja at ikarustech.com<mailto:mutlaqja at ikarustech.com>>
Sent: Friday, April 14, 2017 5:38 AM
To: myxlflik at hotmail.com<mailto:myxlflik at hotmail.com>
Subject: NGC Thumbnails

Hi John,

I sent you invitation to Google hangouts chat as well in order to discuss anything quickly. So I looked at the NGC images and they're great. However, I noticed many of them are _severely_ out of focus or bad quality. Some of them are excellent quality (e.g. thumb-ngc2175.png) but your mileage varies.

Also some images are too saturated (e.g. thumb-ngc45.png). I'm not sure what is causing this, but can you investigate and find out why? Also, how difficult would it be to generate color RGB images from DSS sources? Is there a way to automate this?

I modified KStars so that it also displays the thumb images in the observation planner if the main image is missing. So things are looking good but we need good quality images before we publish them to end users.

Great work!

--
Best Regards,
Jasem Mutlaq

--
Best Regards,
Jasem Mutlaq

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kstars-devel/attachments/20170415/48cc547c/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: OutlookEmoji-😊.png
Type: image/png
Size: 488 bytes
Desc: OutlookEmoji-😊.png
URL: <http://mail.kde.org/pipermail/kstars-devel/attachments/20170415/48cc547c/attachment-0001.png>