« on: September 18, 2012, 02:21:33 AM »
So now we can have L-bags as well! But where's the red ring?
I went back and put the RAW sharpness settings and unsharp masking to zero before converting the file to TIFF and: halos are gone.You can also reduce the halos in unsharp mask by playing with the unsharp radius, e.g. making it much bigger removes background on a larger, smoother scale. You can increase the brilliance of the stars quite a bit by doing this (by reducing the sky haze), but have to take care to not introduce artefacts on large scales, e.g. foreground objects like the mountains in your case (or the dark lanes in the milky way). You can get around this by using masks etc, but I'm starting to get off topic.
Here is a image of the sun reflecting off a gazing globe. Its a 100% crop, but no black ring.The effect is more pronounced the greater the contrast (steep brightness gradient) and smaller the blown image, i.e. stars and hot pixels are ideal to produce dark halos. The sun in your image is more extended and bright also outside the saturated region, so sharpening shouldn't produce as easily visible halos.
This is a straight conversion from raw to jpg in LR 4, all the settings are nominal, no sharpening or NR.Isn't sharpening applied when using nominal settings? (I'm not too familiar with LR4, but other software apply it by default) If you push sharpening using unsharp mask I'm sure you can produce a dark halo around the bright dot in the right hemisphere. Just to illustrate the effect. (looking closer at your image there actually seems to be a dark edge to the white spots [not the sun] - implying some sharpening may have actually taken place?)
But I also will take care next time and won't do 45 shots within 15 minutes!I have taken long-exposure (30s) shots continously for hours (star trails!) without noticing any significant increase in noise, so I'm not sure this is a real problem. It should be very easy to find out though: do you notice a significant noise increase in your last image (when the sensor is "heated up") compared to your first image (with the sensor still at ambient) of that 15 min interval?
When i say it hasn't been processed and what is going on with images then it is still suggested i am lying and that it in fact has been sharpened, it bugs me a little.I was commenting on Pedro's image. In your case it looks like hot pixels with demosaicing artefacts. Are the bright spots in your picture visible at the same pixel positions in other pictures taken at the same time with similar settings? There's always processing going on in producing images, it doesn't need to be the post-processing that introduces problems.
I can't say but what about the fact that your points of light are ovals?Since your question hasn't been addressed yet... the 'ovals' are due to motion blur.
Anyone out there having the same problems while doing night sky?Yes, I happened to take a night shot with stars using a 5D3 yesterday, and I see this precise phenomenom for saturated stars when the default sharpening is used (just tested). When I turn sharpening off, the black halos disappear. I just tried it but am to lazy to produce screenshots for you... just try it yourself. Using unsharp mask as a "sharpening" filter is particularly good at producing black halos.
Replacement lens that currently costs about same as 100-400. At f/5.6 why would anyone want this lens? Help me understand benefit over 100-400?
I had two of these lenses and both were crap.Thanks for contributing with your experience. Your review strikes me as a bit too emotional, exaggerating and contradicting with other reviews for me to take it seriously, in particular since this also is your first post and we haven't had a chance to gauge your experience, skills, language and general attitudes. I cannot say I've seen internet flooded with negative reviews of the lens. To measure barrel distortion you need to carefully align your paper wall with the sensor plane - they have to be absolutely parallel. This doesn't seem to be the case in your example photo, which looks as if tilted slightly downwards. Distortion has previously been measured to be very low, ~1% barrel at 24mm and 0.02% pincusion at 70mm, uniform across the frame. Very few issues with flare and ghosting, even when shooting into the light. Vignetting sure is a problem at f/2.8, but by f/5.6 it's down to half a stop in the corners, and is noticably better than Canon's current offering.
Good luck with your purchase of this lens....Thanks, I probably will get it at some point, and then I will take care to evaluate the lens to check for any problems you mentioned. Maybe Tamron has problem with its QC, but at this point I cannot exclude user error.
You can only win here if you increase the exposure time. In total you have more noise compared to cropping or using a camera with smaller pixels, but you also have more photons if you increase the exposure time to overcompensate for that and increase DR and SNR after downscaling.Yes, you got it right. The point is that you don't lose (much) light by using a TC, compared to cropping. Yes, the absolute noise increases, but the relative noise (noise relative to the signal) decreases as sqrt(number of photons), if photon dominated. To say that the noise increases can be confusing, since it gives the impression that to get the lowest noise images, you expose for as short time as possible. Ideally, a completely dark image has zero photon noise
Well if you're not willing to upload images that you claim show "everything is soft", how are we to know if you're being truthful?We can now see comparisons posted at the-digital-picture.com. I would say the Tamron looks extremely sharp wide open in the centre, but fairly soft in the corners.
I mainly do long exposure star trails, but I thought I'd try some shorter-exposure high-ISO work with the 5D3 to get "frozen" stars with a meteor streak in the the shot. That requires a lot of luck, I know, but I set out none the less with this goal in mind.Nice, I like the colour of the meteor, you don't see too many of these. A method to improve your odds is of course to take many short exposures sequentially. There should be about 1 meteor per minute (for the whole sky), so it shouldn't take too long to capture one in your field of view, on average only a few minutes total exposure time.
2: you lose f-stops if you put on the TC. (The light gets distributed on an area 2 or 4 times as big, leading to 1 or 2 f-stops loss of light density on the sensor.)Light density, yes, but not total light, so it's not so bad. For this reason, you can improve signal to noise and dynamic range with a TC compared to cropping since you can collect more photons from the same scene without saturating. (you could of course also take multiple exposures with the same result with the crop)
The TC does cost you a stop that the crop body doesn't.