Forum:
Q & A
Thread:
Do more pictures make for a sharper panorama?
|
Re: Do more pictures make for a sharper panorama?
Posted: 31 Aug 2010 at 9:59 GMT
|
go to thread
|
DennisS said: I am interested only in the finished product as published on the internet. Each of the following three panoramas were shot with the exact same camera/lens/pano head, processed exactly the same with PTGui, had "Unsharpen Mask" done in Photoshop and used the exact same project file in Pano2VR. The final file sizes are right at 2.3 megs. 6, 12, and 24 are the number of shots taken around. No Zenith or Nadir shots taken here.
Hi Dennis,
Let me know if I get it right.
you just took 24 shots around, then you stitched them in 3 ways.
1. use all 24 images.
2. use 12 images by skipping one image.
3. use 6 images by skipping 3 images.
am I right?
Nick
|
|
alert moderator
|
|
Forum:
Q & A
Thread:
Do more pictures make for a sharper panorama?
|
Re: Do more pictures make for a sharper panorama?
Posted: 31 Aug 2010 at 6:35 GMT
|
go to thread
|
Hi Dennis,
is the pano at max resolution? what is the resolution?
nick
|
|
alert moderator
|
|
Forum:
Q & A
Thread:
Still stitching errors, but located NPP as good as possible
|
Re: Still stitching errors, but located NPP as good as possible
Posted: 30 Aug 2010 at 16:34 GMT
|
go to thread
|
Doug Aurand said: Nick
Which Blender where you using with the Apples & Oranges pictures and Arrow pictures that took the Arrows out and selected either whole Apples or whole Oranges, but didn't blend them?
Doug Aurand
Albuquerque, NM
Just the PTGUI 8.3 blender. Download it and give a try.
Nick
|
|
alert moderator
|
|
Forum:
Q & A
Thread:
Still stitching errors, but located NPP as good as possible
|
|
Forum:
Q & A
Thread:
Still stitching errors, but located NPP as good as possible
|
Re: Still stitching errors, but located NPP as good as possible
Posted: 29 Aug 2010 at 15:55 GMT
|
go to thread
|
Hi Doug,
I spent more time experimenting with ptgui. I first stitched 2 images. Then I placed a few apples and oranges in my source images in such way that they roughly overlapped. They are updated in ptgui automatically. Then I output blended and layered panos to see what happened.
Here are the results.
layered image1
layered image2
layered images overlaid with 50% opacity
full size image of the overlaid
blended pano
blended pano at full size
You see ptgui first find the seam and prioritize the pixels in some magical ways in relation to surrounding pixels and the distance from the center of image. Pixels closer to image center usually have better quality and gain higher priority. Blending of 2 layered pixels is only done in close proximity of a seam.
Nick
|
|
alert moderator
|
|
Forum:
Q & A
Thread:
Still stitching errors, but located NPP as good as possible
|
Re: Still stitching errors, but located NPP as good as possible
Posted: 29 Aug 2010 at 8:46 GMT
updated: 29 Aug 2010 at 8:48 GMT
|
go to thread
|
ok, a real life example of how PTGUI 8 Pro handle overlapping region.
I put a black arrow and red arrow in the overlapping region of 2 images. The location of arrows are outside (further away from image center) the mid point of the overlapping region.
In the stitched image, not a trace of the arrow is shown, indicating that the pixels beyond the seam is not used at all, ie cropped automatically.
This is intuitive otherwise the "blend priority" in ptgui would not be useful.
Gee, I have to reactivate my pbase account just to follow up your argument!
Nick
|
|
alert moderator
|
|
Forum:
Q & A
Thread:
Still stitching errors, but located NPP as good as possible
|
Re: Still stitching errors, but located NPP as good as possible
Posted: 29 Aug 2010 at 7:49 GMT
updated: 29 Aug 2010 at 7:56 GMT
|
go to thread
|
Doug Aurand said: You, Erik & Bigwade claim a stitched image will be better quality if you use 8 shots instead of 6 (Canon 15mm & Nikkor10.5mm) or 6 shots intead of 4 (Sigma 8mm) because you use more of the "best part of the lens."
Hi Doug,
I actually don't like to argue with you because it will be a never ending process. We know that. 
First of all, if I have to choose who to believe, it will be Erik and Bigwade. They have shown their wonderful works. It is what we call credentials, right? They just shared their conclusion from their past experiences. There is no need for them to give any proof. Believe it or not, it is up to you. Well, you want to challenge their conclusion. You on the other hand need to give your counter arguments using valid "real life" examples. I emphasis "real life" examples because they get the conclusion from real life experience. Real life examples will not be affected by any wrong assumptions. So would you spend some time to do a valid real life experiment? You can open a new post to share your finding. I will also do some follow up with real life experiments if I find your conclusion not valid. As Dennis pointed out. Talk is cheap!
secondly, you can't use more of the "best part of the lens." The area of the best part of the lens is fixed. you can't get more of it. Erik and Bigwade get better panoramic image quality because they ONLY use the "best part of the lens." The way how they do this is not important. Maybe their stitchers will discard the excessive part automatically. (I believe latest ptgui blender/ smartblend do so. This is indicated by sharp broken lines in some panos.) Maybe they crop the images before stitching. This just part of their workflow, pure technicality. It will not affect their conclusion that one can get better panoramic image quality by using only the best part of the image circle.
This is the MTF of Canon 15mm fisheye.
how to read MTF chart
www.luminous-landscape.com/tutorials/understandin...
For taking panoramic images, aperture is usually at F8 or above. So just focus on the blue lines.
As can be seen on the chart, the thin blue dotted line (30 LP/mm) drops dramatically at distance beyond 5mm from center.
Now look at the Fisheye Radial Mapping studies by Michel Thoby,
michel.thoby.free.fr/Compare_Fisheyes/Some_plots/...
Canon 15mm fisheye is not studied unfortunately. Let's assume it is similar to Tokina 10-17mm at 15mm.
Typically we shot at 60 deg intervals (ie 30 deg WRT optical axis on the chart), the seam location will correspond to about 7.2mm from image center. Erik suggests 10 shots around, ie 36 deg interval (ie 18 deg WRT optical axis on the chart), the seam location will correspond to about 4.5mm from image center. The contrast/ MTF characteristic of thin blue dotted line drops from 8.2 to 6.5. So Eriks recommendation is perfectly in agreement with MTF chart of Canon 15mm fisheye. Here I have assumed the stitcher discard pixels beyond the seam, ie mid point of overlapping region.
Doug, sorry, I don't present any real life experiments. I will leave this job to you.
Nick
|
|
alert moderator
|
|
Forum:
Q & A
Thread:
Still stitching errors, but located NPP as good as possible
|
Re: Still stitching errors, but located NPP as good as possible
Posted: 27 Aug 2010 at 2:55 GMT
|
go to thread
|
Doug Aurand said: Its common sense. Like if you blend 2 shots of the same scene when one is in focus and the other is a little out of focus, the result will be between the original 2
Doug,
So this is the hypothesis your theory is based on? I am not expert in this field. But there can be infinite ways (mathematical operations) to blend 2 layers of images. You just assume stitchers simply average the layered images. AFAIK, they first find a seam of the overlapped images before blending.
Have you look at what Enblend/ Smartblend can achieve?
This is an example. see how an apple can blend with an organge!
www.cambridgeincolour.com/tutorials/photo-stitchi...
Have you heard of Exposure blending and focus blending? Enfuse, Tufuse?
Nick
|
|
alert moderator
|
|
Forum:
Tips & Tricks
Thread:
New Canon Fisheye
|
Re: New Canon Fisheye
Posted: 26 Aug 2010 at 13:25 GMT
|
go to thread
|
DorinDXN said: Smooth said: OK, you are changing the EXIF data to reflect a supported lens yes? What are you tricking it to be? Seems to do a nice job.
Regards, Smooth 
EF-S 10-22mm f/3.5-4.5
cheers
Dorin
www.livepanoramas.com
Hi Dorin,
I thought lens correction was based on calibration data which was specific to a lens.
Is that just a co-incidence that EF-S 10-22mm has CA similar to Sigma 8mm?
nick
|
|
alert moderator
|
|
Forum:
Q & A
Thread:
Pros & Cons of Nodal Ninja Ultimate?
|
Re: Pros & Cons of Nodal Ninja Ultimate?
Posted: 26 Aug 2010 at 7:52 GMT
|
go to thread
|
Trausti said: Of course precision is the key to success in this field, but it might be a good idea to make plungers available like "extras" for those who feel confident enough to use them. ... and with the quality of stitching software nowadays, a little bit of possible movement would not be a catastrophic thing. After all your heads are very tight and good. I know that I would absolutely go for them. I never looked for those locally or online, but I would go for them if you offered them in your product line.
Trausti
Possible movement/ wobbling has 2 consequences.
1. imprecision in tilt angles.
2. possible blurring in low light situation.
Blurring and extra cost were my major concerns.
The shape of the tilt sockets need to be changed if a plunger is used. So I will keep R1 as is.
nick
|
|
alert moderator
|
|
Forum:
Q & A
Thread:
Pros & Cons of Nodal Ninja Ultimate?
|
Re: Pros & Cons of Nodal Ninja Ultimate?
Posted: 26 Aug 2010 at 7:41 GMT
|
go to thread
|
Smooth said: Yeah but!, you still have to change it from side to side to meet your requirements of degrees of tilt. When it's cold and fingers are numb or wearing gloves it is easily dropped and thus is a con. Even being clumsy it is a small item. Personally I believe it should be tethered.
I don't hear users losing the knob any more after the tweaks. If I receive more request for a tether, I will make a new knob with tether and offer an inexpensive upgrade. Since the knob can be used on both sides of R1, it is best tethered to the rotator at the bottom.
Nick
|
|
alert moderator
|
|
Forum:
Q & A
Thread:
Pros & Cons of Nodal Ninja Ultimate?
|
Re: Pros & Cons of Nodal Ninja Ultimate?
Posted: 26 Aug 2010 at 7:24 GMT
updated: 26 Aug 2010 at 7:25 GMT
|
go to thread
|
Trausti said: My first thoughts were to replace the current locking knob with spring loaded plungers - one on each side - such as this:
www.jwwinco.com/products/section10/lrsp/index.html
It would make it a lot faster to change angles, and with them "permanently" placed, one would not lose them by accident.
Could you possibly consider something like that for a future update Nick?
I actually started with a plunger which can be disabled on the fly. But a plunger (no matter how precise) has moving parts which make the tilt angle less precise. Plus it is more costly to make a precise plunger.
Maybe I can trade precision with convenience in R2. 
Nick
|
|
alert moderator
|
|
Forum:
Q & A
Thread:
Pros & Cons of Nodal Ninja Ultimate?
|
Re: Pros & Cons of Nodal Ninja Ultimate?
Posted: 26 Aug 2010 at 3:08 GMT
|
go to thread
|
DennisS said: Trausti,
Come one. A stock Ultimate pano head does not pivot around the NPP. I really hate to defend Doug here, but he is (this is sticking in my throat right now, kind of tastes like vomit) ...right... That is why additional hardware and/or tripod movement is needed to capture the Zenith or Nadir.
Well, Trausti uses a tripod with built-in tilt head. I don't classify it as an additional hardware. 
So based on Trausti input, with a tilt head and adjustable center column and some skills, R1 can give you perfect zenith and nadir shots.
In fact, one can use view point correction to stitch zenith and nadir shots on R1 with acceptable results.
BTW, I am developing a compact tilt head for R1/10. With an adjustable center column, One can achieve perfect zenith/ nadir shots much more easily.
nick
|
|
alert moderator
|
|
Forum:
Q & A
Thread:
Pros & Cons of Nodal Ninja Ultimate?
|
Re: Pros & Cons of Nodal Ninja Ultimate?
Posted: 26 Aug 2010 at 2:56 GMT
|
go to thread
|
David Tan said:
This is the 360precision absolute ARM that costs about USD 180-200 (shipping included). The arm is different for each pair of camera and lens.
Anyway, maybe I should stick with 360precision absolute for now...waiting for the NN Modular Ultimate next year.
R1/10 is a different beast from 360P Absolute/Adjuste or the NN Ultimate Modular system. It is a single row pano heads dedicated for travelers. It requires fisheye lens with wide enough angle of view. Sigma 8mm is its perfect partner. Its weight and bulk is less than half of 360P Absolute or the NN Ultimate Modular system.
nick
|
|
alert moderator
|
|
Forum:
Q & A
Thread:
Pros & Cons of Nodal Ninja Ultimate?
|
Re: Pros & Cons of Nodal Ninja Ultimate?
Posted: 26 Aug 2010 at 2:45 GMT
|
go to thread
|
Smooth said:
The other obvious flaw with the NN R1/10 is the loose untethered angle degree adjustment pin which is easily dropped and lost. This should be tethered like they are on the RD series rotators.
I made some tweaks to the "tilt angle locking knob" some time ago. Some rubber rings are added to increase the friction between knob and the socket. New knob should not loosen easily or get lost easily. Adding a tether makes it less convenient to use. I will only use it as a last resort.
BTW, R10 has fixed tilt angles. It doesn't have this concern.
Nick
|
|
alert moderator
|
|
Prev
