Karin Posted June 13, 2016 Share Posted June 13, 2016 As a client of ours is deploying quality control of renders, another company has been caught cheating in their 3d-illustrations. We are now asked to describe how we arrive at the representation of the new buildings and document their accuracy compared to the existing surroundings. (This is regarding photomontages with a rendered project composited against a photo background.) Have any of the users here been asked to document the accuracy of you renders?How do you - in that case- do it? (It is a timely, yet strange idea to quality check cg-illustrations, but as they often are delivered as a representation of a "future truth", is is a reasonable question.I just wonder how they can be controlled, as there is so many software packages and different workflows involved in producing this kind of imagery) Link to comment Share on other sites More sharing options...
Tech Posted June 13, 2016 Share Posted June 13, 2016 Hi Karin, What type of accuracy are you trying to document? (Is it the shadows / sun studies, the actual matching of the view with regards to the background image, or something else?) Link to comment Share on other sites More sharing options...
palarsen Posted June 14, 2016 Share Posted June 14, 2016 Send the model in FormZ free or the iOS viewer. Let them look at the model them selves. A render is always "cheating" Link to comment Share on other sites More sharing options...
Karin Posted June 14, 2016 Author Share Posted June 14, 2016 For this client it will be he actual matching of the view with regards to the background image. Link to comment Share on other sites More sharing options...
Tech Posted June 14, 2016 Share Posted June 14, 2016 Hi Karin, Ok, then as far as matching the view will be accurate, this will be as accurate as long as you have built the model correctly, and properly linked the points in the image to the geometry using the Match View tool.... Link to comment Share on other sites More sharing options...
Karin Posted June 14, 2016 Author Share Posted June 14, 2016 The Match view tool has not performed well for us and I stopped trying to use it many years ago. (My former boss advised not to use it, and match the view manually instead)Maybe it is time to give it a second chance. ( I'm still using v6 (because v8 is not performing well on my system either) so I guess there is nothing new...)Still curious; is there is anyone else of the forum users experiencing third party quality control of their images? Link to comment Share on other sites More sharing options...
Tech Posted June 14, 2016 Share Posted June 14, 2016 Karin, The Match View tool in v7 and 8 is far superior to that of v6. Give it a try... Link to comment Share on other sites More sharing options...
rich f Posted June 14, 2016 Share Posted June 14, 2016 You should ask this question on the cgarchitect.com forums. Many Viz professionals there who I am sure would have an opinion. I can't imagine taking on the liability for a legally-binding image. As mentioned earlier, every rendering is a cheat to some degree. As a start I would say you need to model enough of the context as accurately as possible, to be able to flip back and forth between the all-digital version and the photo. So that whoever is making the decision can see that your model elements are locked to the same elements in the photo. Though in my experience, even when you know your camera point and FOV, getting that truly locked in is problematic. (though I also have not tried the match tool in 7/8) Link to comment Share on other sites More sharing options...
palarsen Posted June 14, 2016 Share Posted June 14, 2016 I once had to document accuracy. Camera position and height. GPS and my eye-height. Known far away points in landscape or existing houses. Modelled form map. To correctly adjust the camera viewing angle Horison check. To confirm point of interest for height. If all these match. Your modell is correct in the image. Link to comment Share on other sites More sharing options...
setz Posted June 14, 2016 Share Posted June 14, 2016 I use PhotoModeler for this. Pricey but super accurate. It's a photogrammetry package, I pull geometry, dimensions and camera positions from photos. Link to comment Share on other sites More sharing options...
Des Posted June 15, 2016 Share Posted June 15, 2016 Hi Karin, I regularly have to provide 'Methodology Statements' along with the photomontages. Sometimes local planning authorities request this as further information to a planning application when they want to see the impact of a proposed scheme in it's environment. They, or a planning consultant specify the photo positions from sensitive points of view (from public roads close to other properties whose owners may have issues with the proposed scheme) and which may or may not be visible depending on the view/tree coverage etc. If they don't provide the positions, request them. I usually provide a list of bullet points in paragraph headings such as, "Preliminaries", list of information supplied like the CAD files of the design, specified photos and their GPS locations (even the equipment used). "Computer work", information of the modelling process with the GPS photo positions inserted into a world coordinates file (even the software used). "Photographic matching", the process of matching the model into the photograph. Name the views with the GPS's/Camera exposure etc. "Image Rendering", the process of materials setup, lighting for each photo, camera exposure etc. "Post Processing", the process of inserting the render into the photo using alpha channels etc. You should use terms such as "accurate as possible" and "I believe to be as close as possible to being representative of the design" and "based on current technology" etc. Do not say "this is as it will be" or infer that your image will look "exactly" like this etc. Remember that any 3d modelling/rendering software can only produce images based on the information put into it. The user can only input accurate info during the modelling stage, after that it's interpretation i.e. materials, camera distortion, lighting (to a certain extent) all need to be interpreted by the modeller/user. Even if you know the time, date, position, exposure, fstop, focal length of a photo, you will still need to interpret the cloud cover, diffuse, material reflection, shadow blur etc. Even HDRI lighting needs a certain amount of manipulation. There is a "blind them with science" element to this, so put in as much information as possible of the work done (assuming the work is done well). I have even mentioned other projects and clients in the statement (to show them I'm not bullshitting). Tip; if your scheme is being placed into an area with other buildings, landmarks etc., show them modelled in wireframe/hidden line on a separate image to show the alignment. So, my usual setup is one sheet showing three images; Montage with the render superimposed, Existing photo, Existing photo (maybe faded) with a wireframe/hidden line overlaid to show the position etc. You may also want to provide a map showing the positions of the photos on a separate sheet with a spreadsheet of the views and GPS points. I hope this helps, Des Link to comment Share on other sites More sharing options...
Martin Malinski Posted June 15, 2016 Share Posted June 15, 2016 I have had clients request documentation, though not as detail as GPS positions??....more mapping of general locations, directions and time of day of shots.....It may seem an over the top requirement but when I then consider that there have been clients that use my 3d renders to critique the final build it takes on a whole new dimension of idiocy.....(had some clients trying to match renders accurately to RAL colors also......) Just give them what they ask for.....FormZ is capable of producing accurate photo-montage images.....-why your former boss told you not to use the photo match tool is a little mystifying....?? v6 matching was awkward but still worth the effort, v7 and 8 are much easier. If you were not using the match tool....maybe its the reason the client is questioning the accuracy and requesting proof of representation of images? Link to comment Share on other sites More sharing options...
Karin Posted June 16, 2016 Author Share Posted June 16, 2016 Thank you very much, Des - your answer is very helpful!I am grateful you took the time to give such a detailed and organised reply and will use it as a guidance to set up my own methology statement.I suppose the whole idea is to make the process possible for a third party to recreate and review.It is also good for our company to use as production criteria as we more and more often hire other 3d-artists to do renderings for us. Link to comment Share on other sites More sharing options...
Des Posted June 16, 2016 Share Posted June 16, 2016 You're welcome Karin. You can email me if you want a sample.. Des Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.