I agree with you! The info you have provided is amazingly helpful!
You are also right about proper titles or specialized fori - you always need to find someone with similar interests and willingness to share… With off topic I mean simply in terms of the thread title. Some admins are more strict than others in that respect. And even though I can understand it in principle, most hits probably come from search engines, so not really any harm done!
I will shut up about that now…
So you have a bigger drone too?? For the “real” stuff?
The gimbal you use seems like a very good deal! Is it reliable?
Do you think it could be easily altered to be mounted on a rod?
The base plate would need to come off in any case.
I guess I could just remove the motor at the 4 screws and make a new, smaller one…
What interface do you use to control the movement?
The panorama mount seems like an interesting alternative, although I would also need up and down adjustment.
While this could be achieved by a mechanical system with a string, that would be too much of DIY for my taste…
With RAW processing I meant mainly for photogrammetry.
Because noise reduction is usualy a big no-no, as I understand.
On the other hand, it did work to a reasonably accurate degree even with (as I now realized) completely mangled jpegs…
The stacking method you mention is really interesting, although I doubt it would be practical for larger photogrammetry sets.
About the rod:
I’m sure you know that one: http://www.seaportdigital.com/store/27- … stand.html although I am sure it could be a bit cheaper.
I was contemplating something with a mechanical gimbal bearing on top of a tripod, which I could then mount on a wheeled stand so I can cart it around buildings without much effort. Something like this http://www.geodatasys.com/gimbal.htm only not as gigantic and more stable. The end of the rod would then be weighted so that it stays upright even on slightly uneven terrain. A bit like this: http://www.ebay.de/itm/Jinbei-BM-185A-B … 1246005702 only with a longer rod and bigger base and wheels.
For lightweight, reasonably priced rods, have you looked into supports for portable radio aerials? http://www.qsl.net/dk9sq/dftt.htm The guy is an enthusiast who developed this privately as a hobby. He has a partner in the states - I’m sure you will find the equivalent page. The only thing why I am sceptical is because there are no proper fasteners, so I would be afraid it could collapse (eben though they say it won’t and that people also use it for our purpose). The spiderbeam http://www.spiderbeam.com/index.php?cat … asten.html is more professional, but also more expensive…
Hello Guys.
Tom, in my opinion this type of head isn’t good in photogrammetry. Problem is with “Parallax”. Nodal point is almost in centre. There’s no big difference or no difference in position between key objects on the pictures. Yes 360 Panoramas but not Photogrammetry. I like 360 panoramas and that’s why I bought pano maxx tripod head to eliminate this problem and set nodal point in centre. As I said, it’s just my opinion:]
Götz yes I have:] unfortunately is grounded because Pixhawk 1 Died and now I’m saving money on Pixhawk 2.1 + RTK.
The gimbal you use seems like a very good deal! Is it reliable? - YES
Do you think it could be easily altered to be mounted on a rod? - YES II
“Stacking” if you using Photoshop to automate this process you can use “ACTIONS”, very useful tool:]
Rods - In the year I want to buy one like this 1st one:)
I think Admins should create special topic where we and other users can share experience to improve and optimize all processes.
Robert and Tom, apologies for confusing the two of you!!! :oops:
I guess you can pick whichever was meant for either…
Tom, you are the one who asked about the garden and solar studies, I remember now!
Robert, so did you use de-noised input for RC?
What I meant about RAW stacking is not just the processing but also the sheer amount - if you have a 2000 img model and you shoot 10 burst for each, then its 20.000! RAW! Where should all those files go?!? :shock:
But maybe I just have to get my head around to RAW demands…
I never treated myself to a big Photoshop license - I don’t much care for Adobe products so I only use the cheap small brother which can run actions, but unfortunately not create them… :evil:
Do you know any other software with a similar feature, does Affinity Photo have it?
Because otherwise you would still be stuck with manually processing all those stacks…
I had one problem with interior where I used ISO 1600(no tripod) It was only a few shots. Of course stacking is pointless with big amount of pictures. Mostly I’m using settings ISO 100-200 F/9 so noise is minimal. RAW Therapee and Affinity is good enough.
The method is really great and I am glad you mentioned it!
I know that the standard procedure should be low ISO.
But even with 200, the LX100 shows quite a bit of noise in low light conditions.
Not bad for normal images at all, but just a bit much for my liking in view of photogrammetry.
Using normal denoising did not result in much (if any) loss of detail, actually not worse than the stacking method, which leads to some loss of the sharpness as well, as it seemed to me in the video.
I shall make some tests and report back…
I’m sorry but I should introduce my gear:)
Cameras:
-Nikon D5200 (Main Camera)
-Panasonic LX100
Drones:
-Phantom 4 PRO
-Hexcopter (taort x6 frame)
and PC based on Asus X99e WS motherboard + G SKill 128GB + Asus GF1080 Turbo. Now I’m collecting more hardware because I’m thinking seriously about photogrammetry after when I’ll back to home.
Man, that is serious equipement!
So what’s the LX100 for then?
To take pictures of the pilot?
There isn’t too too much you can upgrade with your PC though.
I think you can quite comfortably start out with that setup.
The D5300 was my choice for a larger camera about a year ago.
We have very similar tastes!
But now I am leaning towards the X T20.
Nice!
Where did you do the rendering?
Or are the highlights in the texture?
The keys are “if” and “know” in this case! :lol:
Clearly I am not quite there yet.
Although my expectations in terms of precision are rather high I guess.
Also, it is quite a difference imho if you are more of a studio- and turntable-guy (camera stationary on tripod) or somebody who is out in the field for hours. I am already getting tired after hours just with the lightweight LX100. That this camera is not even close to otimum is a given! But it does have its advantages, at least to me since I hate clunky equipement…
I have done quite passable reconstructions of architectural features and statues using photos from a Galaxy S7 or even 4k video from the same phone in the past. So for me travelling light is actually an option.
Granted, those are unsuitable for high detail reconstructions due to the rather high amount of sensor noise, but in a pinch they can make your day
Yeah, I guess it just depends what kind of accuracy and detail you are after.
For google earth or somesuch a smartphone camera should be enough any time!
Do you happen to know a program that can raytrace high polygon textured models with a user definable camera standpoint?
I never found one that didn’t cost a fortune or would need a second university degree…
Good old sketchup is too limiting in that respect in my view, but maybe that is due to my old version.
Their prizing philosophy got a bit out of hand I think…
Modern phones cameras are good quality but sensors are still small. In my opinion 12mpx isn’t good if you want to create reconstructions massive objects with many details. Small objects ok, but not big like this coal mine from this post. Dji Phantom 3 Pro has 12mpx(3992x2992) sensor Dji Phantom 4 Pro 20mpx(5464x3640) sensor. Difference between P3P and P4P is massive. I’m not using JPG files because are compressed and you can’t recover details from dark and highlighted areas. Try compare small, tiled, very detailed surface for example bricks. RAW vs JPEG. JPG well be flat. RAW>>TIFF is still detailed. Götz If you want to render high polygon textured models I recommend Blender. Is simple and for free. Of course question is how big objects can lift your hardware? Low poly mesh + normal map mesh is better than 1m polys and of course you can render it in real time engines (unreal, unity). BTW ShadowTail nice model.
ShadowTail, thanks.
That was one I really want to avoid for the reason of the curve. :?
And those two are my friends in that respect, but I already see them so often and sometimes I need to do something that raises the number on my bank statement…
Robert, since you say Blender is easy, I might get back to you on it!
Since I got two recommendations, I will probably look into it sooner than later.
About the resolution:
I would say: Jein! Yes and no (combined and german). In theory, with a smaller resolution you just need to take more images to achieve the same density. That means you have to go in closer which can in some cases also benefit the geometry. Of course, if time is a factor the higher resolution wins! There is an open end anyway, somebody mentioned 42 mpx the other day…
Jpeg we don’t need to discuss, I hang my head in shame that I ever thought I could get away with it! :lol:
No seriously, after all my examples are derived from jpegs - it is possible, but not the end of all possibility. I would argue that in some cases it can still be the better option, depending on the case (there are weird combinations of circumstances out there). I would argue that in very many cases RAW is just being used because it is said to be more professional and the results are in fact worse than somebody else could achieve with a low resolution jpeg. Especially in “normal” photography. My standard documentation I will still shoot in jpeg only because to me hdd space still matters and if you don’t completely mangle them there is no visible difference once they are printed, which I have to for archiving purposes.
Yes, looks good.
The surface seems a little bit rougher than the texture implies, but that could be down to rendering.
What resolution does th 1100D have?
Oh, and how big is the statue in real life?
I attached two of my jpeg atrocities.
And - yes, I admit it - flash, at least partly.
The width of all three is roughly 30cm.
Base:
115 imgages @ 10 mpx
1.5 milion tris
stone-tooling_B.jpg
Capitals: 315/348 imgs and 2.5 million tris
stone-tooling_K.jpg
stone-tooling_K2.jpg
Walls and floor are not always perfect but they are not the POI. On the basis itself you can see which surfaces have been tooled and which ones smoothed, the crudely altered parts are also distinguished. Even the tracing line under the mason’s mark on the base and on top of the second capital come out really well, I think. The bumps are plaster and paint splatters and the funny veins are cobwebs. With the texture of course none of the small imperfections on the walls matter at all. And since my goal was orthophotos, that is even less relevant.
That is very small for so many pixels!
Amazing what render material can do.
I would never have guessed that it is made from sandstone, I though the second render was just shaded…
Did I get it right that your’s are jpeg too?
Shot as such or developed from RAW?
Well, then your first rendering was far more accurate! :lol:
It is a very good imitation though, at least on the pictures.
The cracks and damage is quite realistic, also the fake leftover of vegetation.
So not from the one euro shelf, I guess.
I wondered where you would get hold of such a weathered “artefact”.
Is it covered by sand or just spray painted?
Actually, I remember now why I thought it might be bronze imitation at first, because the lotus and below shows signs of beings sculptured by wax or something - only the shoulder bits look like carved sandstone.
Anyways…
Nice to see that somebody has good experience with jpegs.
With the LX100 the main problem is that it auto corrects a significant distortion at 24mm, as Robert already pointed out.
Other than that I think there would be not much reason to do RAW in my case.
I will definitely experiment and if there is no significant loss, just develop to jpeg.
My suspicion is that the difference will be marginal and only relevant if you have top of the notch lenses and are aiming to squeeze the very last bit out of every pixel. That is necessary for example with aerial photography, where you often can’t just go back and take more images or flight time is so limited, that you have to focus (pun not intended). But in our cases, we can compensate with additional images. In theory, jpeg compression and bit restriction should be evened out a bit like Robert’s stacking technique - the more redundancy you have on one pixel, the more you approach the “real” value by average…
Low poly mesh + normal map mesh is better than 1m polys and of course you can render it in real time engines (unreal, unity). BTW ShadowTail nice model.