The thing is that I don't know how to make it exclude that specific non-stelar structure (for ex the Tadpole's head) when obtaining the starless image - maybe Nikita can help.ĭo you always leave the default settings on? I was wondering if I cannot obtain the same situation as your starting point by creating the stars with starnet, correcting the structures interpreted as stars if any and then subtract the stars from initial image with Pixelmath. So I kind of had to do without, but it is a great tool and you can create perfect star masks. I have to retest it again on some images (I had some great results with starnet for some of my photos), but after the last Pixinsight version it wasn't possible to install the module on the Mac. Of course, you have to give a check on the mask before proceeding on with further tasks I rarely experienced some false star detection with starnet++ on a non-linear image. I saw a post on facebook by Rogelio Bernal Andreo today kind of reminding Juan Conejero that it was an old promise for Pixinsight. That's what's missing in Pix, the layers where you play with opacity and visually dose the quantity of denoise. So if I understand correctly, the denoising is performed with the gaussian blurr on a non-linear image? I have to try that. But since the module doesn't work on Mac anymore in Pixinsight I abandoned it. Starnet can give really good masks though with a bit of tweaking. But I knew it was Starnet since I had the same problem with my processing of the Tadpoles. For example, I saw an APOD submission of the Tadpoles that was published by Sky (the notable images) and one tadpole was missing a head. Never managed to make it work even when changing the step size. I had discovered several times that Starnet mistakes strong signal nebular areas for stars and it can "chop " them off from the starless image. It is really important to check the bright areas. The main one is that Starnet is not 100% proof for differentiating stellar vs non-stellar structures. ![]() I abandoned the idea of working on the starless image and of combining it with the stars later for several reasons. So I try to translate your steps in my head from Photoshop to Pixinsight in order to compare them. I'm a Pixinsight addict so I am not mastering all the Photoshop techniques (I only use Photoshop for the final steps in my workflow). If needed (I mean, if no improvement occurs) I increase the output black levels for the starless image: I blur a little bit the starless and put it on top of my image, as "Lighten" in PS. I create a starless version of my noisy image using starnet++, which is actually a fantastic piece of software. ![]() Hi Mates!I'd like to share with you my denoise and star masking techniques and get some feedback. Use at least 20 flats, 20 flat darks, 20 bias and 20 darks.ħ. Dither when possible to minimize camera defects.Ħ. Sky darkness also comes into the equation for optimal exposure time.Ģ. I usually aim for 6 hours of luminance with my camera, 21.5 SQM sky and 24" mirror. Your optimal subs may be longer or shorter. For the FLI PL16803, this requires at least 15 minute subs. Take as many Luminance subs as possible for the highest stacked SNR for your aperture size, camera type and sky. But you also need to consider reducing noise during acquisition.ġ. I still find I get the best results doing the above. I've tried all sorts of de-noise software and Pixinsight. I then check the final LRGB at the end and repeat the above with very mild fine-tuning on both high SNR areas (positive mask) and low SNR areas (negative mask). When doing RGB, there's another Photoshop control to help minimize color noise. I do this on Luminance and RGB separately. Use Photoshop Camera Raw denoise with adjustments to luminance and detail to smooth noise without affecting the very faintest stars. Play with the levels so there's a smooth mask transition between high and low SNR areas.Ĥ. Stretch the mask using levels so you mask areas having a high SNR. Create a mask layer in Photoshop using the entire image as a mask.ģ. That's right for me.Interesting technique Roberto. In the result there is some lost of detail but I just want to look on my computer screen to 4896x3264 (180 pixels/inch) photos. The small sensor in this kind of camera can't give ultra sharp pictures, so sometimes I use gimp to correct some photos. Lichting effects, smoke, loud bass sound vibrations, moving persons on stage and around you and in this case dark weather make it difficult to capture acceptable photos. For easy manipulation I use a small camera with a 30x optical zoom. Making photos on a music festival is always a challenge.
0 Comments
Leave a Reply. |