Luminous Landscape Forum
Equipment & Techniques => Digital Cameras & Shooting Techniques => Topic started by: Dinarius on October 17, 2018, 09:18:51 am
-
I use my Canon 45mm Tilt-Shift a lot.
Occasionally, I'm in a situation where the lens isn't wide enough for an architectural view. Usually, in these situations, I use a wide (non tilt-shift) lens and fix the verticals afterwards in Capture One.
What I'd like to know is this: if I want to shoot a flat building facade that is too wide for my 45mm, should I pan and stitch, or should I shift left and right and then stitch?
I've tried panning in portrait format, allowing about 2/3 overlap, but I'm not happy with what I'm getting from CS6 afterwards. I'm careful about keeping everything parallel and vertical while panning, but I still don't like it.
What is the best way to get a wide building flat and upright, without hiring a 24mm T/S! :)
Thanks.
D.
Ps. If this is the wrong forum, please move.
-
There will be people more experienced than me chiming in, but I would venture to say that a shifted-then-stitched image would be the way to go, with more natural look and less distortion. I think (just think) that in order to achieve the same result with a panoramic photo (I assume that is what you meant by "pan") you would need to know exactly where the entrance pupil (a.k.a. nodal point) is and use a special panoramic contraption.
I captured this using Canon's 24mm T/S, Mark II, using shifting and stitching:
-
There will be people more experienced than me chiming in, but I would venture to say that a shifted-then-stitched image would be the way to go, with more natural look and less distortion. I think (just think) that in order to achieve the same result with a panoramic photo (I assume that is what you meant by "pan") you would need to know exactly where the entrance pupil (a.k.a. nodal point) is and use a special panoramic contraption.
I captured this using Canon's 24mm T/S, Mark II, using shifting and stitching:
Slobodan,
Thanks for the reply.
My problem with shift and stitch is that I usually use vertical shift to take in the upper stories of a building and remove unwanted ground.
If I use vertical shift, then I can’t use lateral shift to create a multi image shot.
Or am I missing something?
Thanks.
D.
-
Actually, you can, but you will be losing comers. I have an example, will post later, when I find it.
-
Found it: https://forum.luminous-landscape.com/index.php?topic=111108.msg916012#msg916012
-
I use my Canon 45mm Tilt-Shift a lot.
Occasionally, I'm in a situation where the lens isn't wide enough for an architectural view. Usually, in these situations, I use a wide (non tilt-shift) lens and fix the verticals afterwards in Capture One.
What I'd like to know is this: if I want to shoot a flat building facade that is too wide for my 45mm, should I pan and stitch, or should I shift left and right and then stitch?
I've tried panning in portrait format, allowing about 2/3 overlap, but I'm not happy with what I'm getting from CS6 afterwards. I'm careful about keeping everything parallel and vertical while panning, but I still don't like it.
What is the best way to get a wide building flat and upright, without hiring a 24mm T/S! :)
Thanks.
D.
Ps. If this is the wrong forum, please move.
Maybe it was good an idea to use both techniques at the same time. I sometimes do.
If the building is too high and broad to fit into my picture, I try horizontal shifting the camera in vertical position and same time tilting the camera upwards.
Then I stitch shifted shots in Photoshop (CS6) and straighten vertical lines, which is often easy in architectural shots.
When correcting verticals in post you loose some sharpness, but with shift-stitch technique you get more pixels into picture and can possibly downsample it to get
better final photo.
-
I use my Canon 45mm Tilt-Shift a lot.
Occasionally, I'm in a situation where the lens isn't wide enough for an architectural view. Usually, in these situations, I use a wide (non tilt-shift) lens and fix the verticals afterwards in Capture One.
What I'd like to know is this: if I want to shoot a flat building facade that is too wide for my 45mm, should I pan and stitch, or should I shift left and right and then stitch?
I've tried panning in portrait format, allowing about 2/3 overlap, but I'm not happy with what I'm getting from CS6 afterwards. I'm careful about keeping everything parallel and vertical while panning, but I still don't like it.
What is the best way to get a wide building flat and upright, without hiring a 24mm T/S! :)
Hi D.,
The highest quality would be delivered by a 'Pan and Stitch', because you're shooting with the center of the image circle. The problem with CS6 is that it offers too little control, so one can quickly run into limitations of the Stitching module. If you were to use a better Pano Stitcher, e.g. PTGUI, you would have all the required controls (even for shifted image tiles), and it allows to compensate for keystone distortions, and rotations, and it removes any residual lens distortions (no need for separate distortion correction per image tile).
Left and Right shifting, if used with an entrance pupil correction while shooting, is the simplest, because the files only need to be shifted in alignment in Photoshop, not warped. But shifting the lens means that the edges of the image circle are used on the L/R edges of the composite.
Cheers,
Bart
-
Slobodan,
Thanks for the reply.
My problem with shift and stitch is that I usually use vertical shift to take in the upper stories of a building and remove unwanted ground.
Photoshop doesn't like that. PTGUI can handle that just fine.
If I use vertical shift, then I can’t use lateral shift to create a multi image shot.
In that case, you could try and shift 45 and 135 degrees up if that allows a wide enough FoV (also vertically). But Panning is easier and only requires a minimum of hardware for the alignment of the entrance pupil of the lens. You can pan with a shifted lens, but you can also correct for keystoning in the Panostitcher (PTGUI understands shifted tiles and can stitch them, but it can also correct keystoning afterwards on unshifted tiles).
Cheers,
Bart
-
My gut says you should do one method or the other, but using both may muddy your results. But I'm no expert on T/S since I have never used one.
When needed I always pan, stitch, and correct for architecture. I leave some extra room if I can to give me some "slack" I can sacrifice for corrections.
I have found Lightroom's Upright feature to work very well most of the time (but not all) for straightening this kind of distortion.
-
Thanks for all the replies.
On a shoot at the weekend I had shot a few horizontal and vertical panoramas using the 45mm T/S, as well a shooting the same building with a wide lens tilted up, just to cover myself.
In the end, having tried CS6, I used a single shot from the wide lens and just corrected the verticals. I had tons of file for what I needed.
However, I just downloaded the PTGUI trial (thanks Bart!) and processed a three shot, landscape-format, panorama (where I had shifted the 45mm up a bit to take in the top of the building and some sky, and lose some of the lower foreground and then panned three exposures) and the result is flawless. One big gorgeous file.
In order to have as little as possible to do in CS6 after creating the panorama, I processed the files in C1 and then saved them as TIFFs.
1. Is this what others would do? There is no option to create a RAW format panorama; correct?
2. I watched the PTGUI demos and I cannot get my head around the positioning of the lens and the use of that tripod rail gadget to keep it in the same position. Any better video you can point me to? (I had also shot a portrait-format six-shot panorama of the same building and I suspect that this would have benefited from keeping the lens in the right position.)
3. Any other PTGUI workflow suggestions?
For what I do, using PTGUI would be a lot cheaper than hiring (or buying) a 24mm T/S, which I do occasionally.
Thanks.
D.
Ps. This is the head I'm currently using: https://www.manfrotto.ie/405-geared-tripod-head-strong-and-lightweight-aluminium
-
However, I just downloaded the PTGUI trial (thanks Bart!) and processed a three shot, landscape-format, panorama (where I had shifted the 45mm up a bit to take in the top of the building and some sky, and lose some of the lower foreground and then panned three exposures) and the result is flawless. One big gorgeous file.
In order to have as little as possible to do in CS6 after creating the panorama, I processed the files in C1 and then saved them as TIFFs.
Yes, that would produce excellent input tiles for PTGUI. Stellar Raw conversion quality, Chromatic Aberration correction, Diffraction Correction, and if preferred Structure and Clarity in addition to Color and Tonal corrections.
1. Is this what others would do? There is no option to create a RAW format panorama; correct?
I would, and no, neither CaptureOne nor PTGUI offers to save to a Raw (because it's not Raw anymore).
2. I watched the PTGUI demos and I cannot get my head around the positioning of the lens and the use of that tripod rail gadget to keep it in the same position. Any better video you can point me to? (I had also shot a portrait-format six-shot panorama of the same building and I suspect that this would have benefited from keeping the lens in the right position.)
The most benefit of keeping the entrance pupil in a stationary position comes from total freedom when adding tiles to expand the FoV. For facades with little detail in the foreground (street furniture, trees/(lamp)posts/etc.), alignment to keep the entrance pupil stationary is less critical. The more depth there is in the image, the more critical alignment becomes.
One can use gear to optimize the setup for either single row, or multi-row stitches.
- Single Row: If the optical axis of the lens does not align with the horizontal center of rotation, one will first need a horizontal bar for shifting the optical axis into alignment such that it intersects with that rotation center.
- Single Row: Requires a minimum of gear, usually only a bar that allows shifting the camera/lens forward or aft, to align the vertical axis through which one rotates with the entrance pupil of the lens. That works if the lens is already aligned with the horizontal tripod/rotating head offset. This then also allows using a vertically shifted lens and only horizontal panning.
- Multi-Row: If one also wants to tilt the lens axis up/down, then that rotation also needs centering on the axis through the entrance pupil, by vertical shifting on a bar. That will add some complexity to the setup.
So to summarize; single row stitching requires alignment of the entrance pupil with 2 axes (horizontal shift and fore/aft shift aligns the vertical axis of rotation), and multirow stitching requires alignment of 3 axes, to keep the intersection of 3 axes aligned with the center of the entrance pupil.
3. Any other PTGUI workflow suggestions?
Meticulous leveling on tripod is not necessary anymore (approx. horizontal leveling does help because it results in less cropping / content aware filling), so there's a lot of time saving and no requirement for high quality (electronic) levels (which also need calibration). Leveling can be a part of the stitching process and can be helped by manually adding vertical control points (that are often available with architecture). Horizontal control points only make sense when shooting exactly perpendicular to the front of the facade. Once keystoning is minimized, it can help to deliberately add back a little keystoning by dialing a small amount of pitch, especially on shots from a relatively short distance (that get stretched vertically a lot).
Saving the project file (and the tiles, or the Raw converter parameters) will allow recreating a new stitch at a different size (even from a somewhat different angle). PTGUI uses excellent resampling algorithms that preserve lots of detail despite the warping that takes place.
For what I do, using PTGUI would be a lot cheaper than hiring (or buying) a 24mm T/S, which I do occasionally.
Absolutely. It can even replace an expensive Tilt and Shift lens with a very high quality regular fixed focus length lens. On a 36x24mm full frame sensor, an approx. 50 mm lens offers a very good compromise between lots of magnification of detail (compared to a wide-angle) and still enough depth of field.
Cheers,
Bart
P.S. I also have the Manfrotto 405, but use it more for product shots and macro work. For Panos / Architecture, I use a lighter weight (but a just as sturdy combination of an EZ-leveler II and an indexing rotator on top of that, with an RRS clamp and bars on top of that). Relatively lightweight and easy to transport compactly.
-
Bart,
Thanks again. Much to take in - I’ll be back to you.
Meanwhile, Helicon Focus (excellent software) will take any number of RAWs and give you back a .dng.
Would be nice if PTGUI could do likewise.
Thanks.
D.
-
Bart,
Thanks again. Much to take in - I’ll be back to you.
You're welcome.
Meanwhile, Helicon Focus (excellent software) will take any number of RAWs and give you back a .dng.
Would be nice if PTGUI could do likewise.
I think that one of the issues is that DNGs are a variation of a TIFF. Focus-stacking will produce a file that has approx. the same pixel dimensions of the source images. Panos, on the other hand, can easily exceed the 4GB TIFF file limit. PTGUI, therefore, does support PSB file output (e.g. for Gigapixel panos).
Cheers,
Bart
-
Lightroom 6 and newer produces dng files for either hdr or panoramas.
-
Lightroom 6 and newer produces dng files for either hdr or panoramas.
Hi Matt,
I'm not sure if Adobe has fixed it by now, but the LR Version 6 Pano's didn't allow to remove Chromatic Aberrations in the individual tiles of such composite "Raws", effectively making it impossible to remove it afterward in Panos. Besides, I believe there is a dimension limit in Lightroom of 65000 pixels maximum, which can be a real limitation for wide Panos (and I don't know if it has the same 4GB total filesize limit as TIFFs).
Cheers,
Bart
-
Found it: https://forum.luminous-landscape.com/index.php?topic=111108.msg916012#msg916012
Slobodan,
Thanks for that.
D.
-
The most benefit of keeping the entrance pupil in a stationary position comes from total freedom when adding tiles to expand the FoV. For facades with little detail in the foreground (street furniture, trees/(lamp)posts/etc.), alignment to keep the entrance pupil stationary is less critical. The more depth there is in the image, the more critical alignment becomes.
Bart,
What is the difference between entrance pupil and nodal point?
Must both be used when shooting panos?
A lot of what I'm doing in an upcoming project will simply involve capturing buildings that would fit fine into a 24mm T/S, but which I'll want to capture with the 45mm T/S to keep the budget down.
As I wrote above, a three-shot pan of a building I did the other day worked fine in PTGUI. That was done without having a head to correct for the nodal point - there was nothing in the near ground that would have been affected anyway.
Am looking at one of these (https://www.manfrotto.ie/panoramic-rotation-unit) in combination with one of these (https://www.manfrotto.ie/panoramic-head)
But, do I need either or both for the exteriors I'll be doing?
Thanks.
D.
-
I'm sure Bart will chime in but the entrance pupil and nodal point are different and not at the same point of the lens. You only should concern yourself with the entrance pupil.
The Manfrotto examples you linked to are very cumbersome - I owned both. Take a look at what is available at RRS.
When I pan (usually single row) I don't concern myself with any special equipment except a short rail to move the camera forward or backward to the entrance pupil. Lots of times I just eyeball in the finder for the next panned shot. Its always best to be a little wider and crop the final image.
Victor
Edit: You can find a lot of useful information at https://www.panoramic-photo-guide.com/finding-the-nodal-point.html
-
Bart,
What is the difference between entrance pupil and nodal point?
The technical difference only matters for optics designers (https://en.wikipedia.org/wiki/Entrance_pupil and https://en.wikipedia.org/wiki/Cardinal_point_(optics)#Nodal_points), for Panoshooters we use the Entrance Pupil or No-Parallax point (the single apparent position of the aperture when looking in the lens from the subject side). The entrance pupil can shift a bit foreward or backward with focus distance, depending on optical design. For architecture, the distance is usuall large enough to allow a single No-Parallax point setting, for interiors if can help to calibrate for a closer focus distance, typical for the size of interior.
PTGUI is very capable in solving small inaccuracies, but mathematically can only achieve zero parallax in a single plane unless the No-Parallax point was used.
Must both be used when shooting panos?
Only Entrance Pupil / No-Parallax point matters.
A lot of what I'm doing in an upcoming project will simply involve capturing buildings that would fit fine into a 24mm T/S, but which I'll want to capture with the 45mm T/S to keep the budget down.
Totally doable, with minimal additional postprocessing, and with great resolution. You can choose whether to use multiple Shifted lens shots or Unshifted tilted shots, all perspective issues can be solved by the Stitching application and, at a distance, any parallax caused by unaligned entrance pupil positions will be small. You will want to nail the alignment of the rotation axis through the entrance pupil with subjects that have a lot of depth, or occlusions like looking through gates/arches or with street furniture, or brick roads.
As I wrote above, a three-shot pan of a building I did the other day worked fine in PTGUI. That was done without having a head to correct for the nodal point - there was nothing in the near ground that would have been affected anyway.
That's key, nothing nearby, and PTGUI by default automatically corrects for the decentered/shifted optical axis. For such a single row setup there is some leeway.
Am looking at one of these (https://www.manfrotto.ie/panoramic-rotation-unit) in combination with one of these (https://www.manfrotto.ie/panoramic-head)
But, do I need either or both for the exteriors I'll be doing?
I also use the 300N Panoramic Rotation Unit, because the indexed stops allow me to get consistent overlaps, and I now rarely skip a shot in large stitching projects by getting distracted with many shots per row. It also allows faster shooting without having to look through the viewfinder. That can help when clouds are racing by due to high wind speeds (I shoot in the opposite direction of cloud travel, to avoid repeating ghosts anyway, so sometimes left-to-right, sometimes right-to-left). The indexing click-stops make life a lot easier, and also allow to shoot with the camera close to a back wall (often the case with interiors), because there is no need to look through the viewfinder while rotating.
I do not like the 303 Panoramic Head, too heavy and the vertical support plate cannot be shifted vertically (which might be required for some cameras if you want to expand to a multirow setup). So it only allows a single row approach and weighs a lot.
Personally I invested in a more modular system based on Really Right Stuff components (http://www.reallyrightstuff.com/RRS-pano-gear) that can be expanded over time as the requirements grow.
For a single Row setup (assuming the optical axis is aligned sideways with the vertical panning (Yaw) axis), the "Package For Shooting Single-Row Panos: Pano Elements, LR" suffices. A camera model specific L-bracket is required.
My multirow setup was assembled over time and the individual components can be reassembled into all sorts of setups (see attached). For my Fisheye lens for 360-degrees VR shots I use a shorter MPR rail with an integrated clamp instead of the longer MPR II, to keep the rail out of the lens' 180 FoV.
EDIT: I've also added the Single Row configuration, which assumes that the optical axis aligns with the center of rotation along the Yaw axis. As you can see, simple, lightweight and quite compact. And when used with a vertically shifted TS lens, the EZ leveler II allows enough adjustment to also shoot some higher buildings without the need for a Ballhead or 3-axis tripod head with a rotator on top.
Cheers,
Bart
-
The discussion shift or stitch is intersting.
I think i depends a lot on the quality of the lenses.
TS lenses are usually not that good when fully shifted, so at some point a stitch will give better results.
In this case the canon 45TS is an old design - my guess is that a stitch with a good 50mm lens ( Sigma ART par example) will give better results.
on the other hand the new 50mm TS from Canon might be better again.
If i make an 19mm photo by stitching with a 24mm lens and compare it with the very good 19mm PCE from Nikon- the PCE wins.
-
Bart,
Thanks again.
RRS not easily available this side of the pond. Purchase and servicing of Manfrotto much easier.
Will see if I can manage with my 405 and PTGUI for the moment.
D.
-
Bart,
Thanks again.
RRS not easily available this side of the pond. Purchase and servicing of Manfrotto much easier.
Will see if I can manage with my 405 and PTGUI for the moment.
D.
RRS is expensive, but the center sensor-marks on the L-bracket are very precise.
Manfrotto does not offer this quality but is alot cheaper.
I happen to have about the same system as Bart.
-
Nodal Ninja is another panorama rig to look at. Well made, well priced.
I've tried the multi-row stitch method to shoot a few facades before and it works fairly well but find it can often look a little strange, kind of like superwide distortion but more that the image has been stretched. Depends how close you are to your subject. I've done it with a 50 and 135 and the 50 works better.
I would generally prefer the same sort of method as Siobodan, shifting the ts-e along one of its diagonal axis, taking 4-5 (8-9) shots (four corners one center as an anchor) and then stitching. I do this a lot with the 24mm to get around a 27mm field of view. The 45 will probably give about a 28-30mm this way. If you rotate the lens off its horizontal shift axis by the first stop, 30degs, you'll get a slightly aspect similar to 16:10. There is also a much more natural look to the final image than the other method.
You could possibly combine the two.
Lightroom will stich files into a dng and does a decent job. It's also pretty good at getting rid of chromatics in the DNG after it's been stitched, but I did find C1 better when you import this DNG. It might not totally clear them up though if there are some heavy ones.
-
Returning to this topic after a pause...
Was on a shoot this week - small building exteriors - and shot a few simple panos. (I also shot with a wide lens to cover myself.)
The pano were 3/4/5 shot setups either portrait or landscape.
In each case (only 4 in total) my CS6 produced a better result than PTGUI.
CS6 result was better in every respect; colour, contrast and perspective correction.
By comparison, PTGUI were flat, requiring a lot of editing. Also, in every case, vertical perspective correction was required. CS6 was pertin this respect.
I accept that I’m a total greenhorn with PTGUI. This is just an observation.
But, PTGUI is €200. That’s 20 months of CS sub, which I presume is an improvement on CS6?
I’m guessing that for complicated stuff PTGUI wins.
D.
-
What kinds of files where you feeding PTGui? It shouldn't be making changes to colour and contrast.
-
Returning to this topic after a pause...
Was on a shoot this week - small building exteriors - and shot a few simple panos. (I also shot with a wide lens to cover myself.)
The pano were 3/4/5 shot setups either portrait or landscape.
In each case (only 4 in total) my CS6 produced a better result than PTGUI.
CS6 result was better in every respect; colour, contrast and perspective correction.
By comparison, PTGUI were flat, requiring a lot of editing.
Hi D.,
Which I suspect, is due to not feeding PTGUI color correct TIFFs. I assume that the reason for my assumption is because PTGUI does no adjustment in those aspects, unless specifically directed to take a guess.
Also, in every case, vertical perspective correction was required. CS6 was pertin this respect.
I accept that I’m a total greenhorn with PTGUI. This is just an observation.
If your "vertical perspective" means keystone correction, then there are several ways to address that, some are quick, some more accurate (assuming that is what it should be, instead of pleasing).
Cheers,
Bart
-
Bart,
Thanks.
A few points.
1. When I input a TIFF into PTGUI (vs. CS6 or Microsoft's offering) it is rendered as if I had selected Linear Mode in CaptureOne. i.e. it appears flat (devoid of contrast or curve) and at least a stop below where it was. But, this could just me my naivety with the software.
2. Images inputted into PTGUI are coming out in need of more keystone correction than the same images from CS6. When this happens in CS6, it's usually easy to use Content Aware Fill to add extra sky (for example) before fixing the problem. Is there a way around this in PTGUI, or do you have to allow more headroom at the capture stage, by default? So far, I've saved the image from PTGUI and fixed it in CS6. Not ideal.
Thanks.
D.
-
Bart,
Thanks.
A few points.
1. When I input a TIFF into PTGUI (vs. CS6 or Microsoft's offering) it is rendered as if I had selected Linear Mode in CaptureOne. i.e. it appears flat (devoid of contrast or curve) and at least a stop below where it was. But, this could just me my naivety with the software.
Switch to Advanced settings if you haven't already, then check the settings on the Exposure / HDR settings Tab (and if needed, Reset Automatic Exposure and Color adjustment). Also make sure you haven't checked the "Apply Tonemapping" box in the Panorama Editor (CTRL+E).
2. Images inputted into PTGUI are coming out in need of more keystone correction than the same images from CS6. When this happens in CS6, it's usually easy to use Content Aware Fill to add extra sky (for example) before fixing the problem. Is there a way around this in PTGUI, or do you have to allow more headroom at the capture stage, by default? So far, I've saved the image from PTGUI and fixed it in CS6. Not ideal.
Not sure if you mean that Keystoning is under-corrected or that the projection is different. If the former, there are several possibilities to either automatically or manually adjust that. If the latter, you can try a different projection in the Panorama Editor (CTRL+E). It may be useful to shoot an additional partial image (row) at the Top/Bottom Center region, and use your Photo Editor to Heal or InPaint the missing parts of an image. A Cylindrical Projection, if shot level, will have equal vertical coverage for each image tile in the Panorama but all horizontal lines (except the Horizon) will be curved (which can be fine for Nature shots, but not Architecture).
Cheers,
Bart
-
I use my Canon 45mm Tilt-Shift a lot.
What is the best way to get a wide building flat and upright, without hiring a 24mm T/S! :)
None. Get the 24TSE. It is a fabulous lens and I now use it as my main landscape lens on the X1D which makes it even wider.
-
Rather than start a new topic, I figured it would be useful to keep everything here.
I shot a pano from a ship of a flat calm sea.
The pano (five frames) is pretty good and most people would have to have the few visible seams between files pointed out to them.
How can I best clean up these few flaws?
I presume that given the nature of the sea, there is no avoiding this problem in future?
Thanks.
D.
-
Rather than start a new topic, I figured it would be useful to keep everything here.
I shot a pano from a ship of a flat calm sea.
The pano (five frames) is pretty good and most people would have to have the few visible seams between files pointed out to them.
How can I best clean up these few flaws?
I presume that given the nature of the sea, there is no avoiding this problem in future?
Hi,
Depending on the nature of the flaws (foreground/background, occlusions, movement, etc.), ultimately a warp or clone/heal tool on the separate layers may be needed.
Cheers,
Bart
-
1. When I input a TIFF into PTGUI (vs. CS6 or Microsoft's offering) it is rendered as if I had selected Linear Mode in CaptureOne. i.e. it appears flat (devoid of contrast or curve) and at least a stop below where it was. But, this could just me my naivety with the software.
Could it be that the TIFFs are saved in Prophoto profile, rather than in RGB?
-
Hi,
Depending on the nature of the flaws (foreground/background, occlusions, movement, etc.), ultimately a warp or clone/heal tool on the separate layers may be needed.
Cheers,
Bart
Bart,
Thanks.
What is visible is the lines between some of the files in the foreground area. Obviously, the constantly changing nature, even of a flat calm sea, caused this. (The background sea and the entire overcast sky are perfect.)
The easiest way of dealing with these would be the likes of the Clone Tool in Photoshop.
The reason I ask the question is that I'm wondering if I can tweak certain settings in PTGUI to make it render a better result.
D.
-
The reason I ask the question is that I'm wondering if I can tweak certain settings in PTGUI to make it render a better result.
Hi,
You could try, with the same overall settings, to do an additional stitch that you can blend in as a layer in Photoshop or a similar editor.
This additional stitch can be made with a different "feather" setting on the last "Create Panorama" tab, or use a dedicated stitch based on foreground controlpoints, and another based on background controlpoints. By forcing PTGUI to use only controlpoints in a plane at a given foreground distance from the camera, a stitch can always be made, but then the parallax will be in the background. Then combine the good parts from the two versions in postprocessing.
If you use the PTGUI Pro version, you also have the ability to use masking during the stitching process, so you can include/exclude features from one tile in the overlap zone.
Cheers,
Bart
-
Cheers Bart,
That's Dutch to me.
But, I'll try and work on it when I have more time.
Meanwhile, if you have a link to a video of what you describe, that would be great.
Thanks!
Denis