I don't know if this subject really belongs here, but I thought it is worth to try, feel free to remove, if this goes too far from what board is for. There are numerous tutorials about this subject that can be find with your favorite search engine, so better use those to find out how, I believe. However there is one issue that I'm experiencing and I would think that some modellers might be familiar with cause and reason why. I can't model very well, I know basics, but results are not so great, but I have been fortunate enough to have modellers to work with quite some time. Now I have found that model that I know have been taken great care to model as perfectly as possible is not matching to photos I take from the car. I took photos from far with DSLR using long range lens, to minimize distortions, I did use middle range of 'zoom' so that there would be even less distortions. Still I find that car in photo is higher than in model. mostly lower part of car seem to reach lower, even I took photos from same height as centerline of side of car is, which should help to minimize the issue. Maybe I should take 3 photos from 3 different heights to further minimize the effect. However when I look car and the photo they look same, but when I look model it does not look enough high, but when I measure it is enough high, only it does not quite reach as low as it should. Now this is not happening just with one car, but there are other models too where I find similar issue. What I'm curious to know is that has others experienced similar things when measuring vehicle from ground level to bottom of vehicle in 3d modelling software? How about when taking photo and putting it to plane and comparing model to photo in side view? With my Volvo model there is visual difference and when I measure real car and use planes to measure distance to ground from model, I get different results, can there be other reason than inaccuracy in model? Also what kind of accuracy one should except or try to achieve when making model, lower part of the car certainly is not most visible one, but should one try to get to 1mm accuracy or is it just plain stupid?
Well...first of all I think you can't search for 1mm accuracy for the bottom distance...as you can get different results with different car settings, loads, pressures etc. About the difference between a picture and a viewport we can't just use a tele lens to get a match. When I work with 3D integrations for my job (Real object in 3D environments or 3D object into photographs), I need to know everything about the picture. The camera model, the focal lens, the CCD/CMOS size, the position, the tilt/pan angles etc..and time and location (if I have to render a GI set) Some of those data are into EXIF, for angles and position you've to work with triangulations, or write down angle/position values when you're the photographer.. When you have all these data in your pocket you can program a 3D camera with same parameters (sometimes you've to convert some parameters, depends on the software you're using), and create a virtual set with same camera distance, same angles, same FOV etc... Of course best shots are by tele lens, with less barrel and aberration due the lens shape...so if you get a proper shot with a normal/tele lens and you know some data you can do a proper comparison. So, do not worry to much about these incoherence between photographs and 3D. If you're working on blueprint (that aren't lens based but ORTHO projections in CAD), you can stay sure you're working on the right side...
That is interesting, what I recon is that I could do panorama with Hugin and choose Ortho projection, 6 shots, maybe 9 would be needed to be taken, but I imagine with that I could get it very close. This is with 3 shots and rectilinear projection, first one plane at middle of car, 2nd one plane is moved towards viewer to more clearly see differences/similarities: View attachment 3286 View attachment 3287 Of course it is just a mockup, car models are different too, there is difference in wheel base in reality so I should take time and get photo and plane accurately to proper dimensions, but in this I was mostly interested from the front section that seem to be relatively close. also distance from rear wheel center to rear seat edge is same in both models and rear seat seems to align quite well, so I believe there is perhaps a way. Just tele lens is not enough, but tele lens + hugin? I need to input my lens data and crop factor to Hugin so it does some corrections, then with several photos and proper projection I believe I should end up very close. Of course modelling itself, I doubt that it could be done to 1mm precision, at least any blueprints I have found become very blurry when I zoom in and try to get precise position, so it always is estimate, unless I measure real thing and move vertex exact amount and repeat for few thousand times Photos should be taken straight head on and at level, no any angle at all, any angle makes problems, so some measurements are needed for proper photographing, I did took shots just from freehand, so there are lot of errors, but still, quite bit closer than without using panorama tool like Hugin (free and incredible good software). I have never managed to get past front fender when attempting model the car, this model is made by MJ and I can only dream to make something as good some day, but it is learning to see, I guess, nothing more.
