By rogersalph
#361961
Hi Guys,
I am trying to find an optimum work flow to do photomontages. Please find in the image below: a quick attempt to match the perspective of the building onto an existing site photo using Match Perspective Command in Sketch Up.

http://minorityexpress.wordpress.com/20 ... sketch-up/

In the image above, I have tried to match the points of existing roof and all context in the model is accurately modeled according to surveyor's drawings. So there is no error in terms of scale relative to design, etc.

The picture above is taken with an EOS SLR 350D, with a camera lens of 20mm. (Factoring in the Crop Factor of 1.6)
I have often found that the quickest way is to distort the design to fit onto the site or trying to manually adjust the photo to fit on site. See below.

http://minorityexpress.wordpress.com/20 ... -on-photo/

However, in tight deadlines and constant change in design, I find that I end up spending hours on image editing, etc. Is there a workflow which you guys have find more suitable in this instance? For example: Setting up a viewport which fits exactly to the image you produce which minimizes the work done on photoshop. I would love to hear from the ways which you guys contextualize your photos. Thanks!
User avatar
By Mihai
#362040
It will likely not be a perfect match if you don't also take into account the lenses distortion. If you import your photo into Lightroom for example it has a database of lenses and knows from the EXIF which lens took the photo and it can automatically undistort the photo based on the lens distortion profile. Some lenses have simple barrel/pincushion distortion, others have more complicated distortion profiles. The more wideangle lens, the more heavy probably is the distortion.

Also in Maxwell, just enter the cameras sensor size and the real focal length of the lens - no need to manually figure out the corrected focal length.

So:
1. Import photo in LW (or other photo apps which can undistort based on a lens profile)
2. Undistort it
3. Export this undistorted one
4. Use it in Sketchup with the photo match tool thing
5. Set the Maxwell camera to the sensor size of your camera and real focal length of the lens
6. Render (with alpha if you want to later comp into the real, undistorted photo)
#362101
Hi Mihai,
Thanks for replying.
I've got two question here.
1. When you say un-distort image. Do you mean that the software can automatically correct the leaning vertical lines of the image. This will be a bit troublesome as I intend to distort the image as little as possible as we are trying to make a case to the local authorities. By manipulating the image, I fear that we will be held accountable as we did it on purpose to prove a particular point.
2. Also, I do realize that whenever I set a FL in Sketchup, the conversion seems to be altered when it is exported to Maxwell. For instance, if I use a FL 24mm in sketchup, it will appear in maxwell as 19.4. Should I be correcting the FL to 24 in maxwell too?


Any comments regarding your workflows and the answer to the above two questions would be great!. Thanks.
User avatar
By Mihai
#362108
1. Not leaning lines, it won't try to straighten a photo you took pointing the camera upwards - but instead small and not so small distortions due to the lens, like straight lines appearing to bulge.

http://www.andromeda.com/people/ddyer/p ... -large.jpg

2. I don't know if this depends on the way you are rendering the viewport - so the plugin tries to match the exact view or partial view by changing the FL. If the ratio of the viewport is not the same as the ratio of a certain film width/height it has to do that otherwise your SU view won't match with the render. I suggest to first set your render area to have the same ratio as the sensor size on your camera. Yours is 22.2 x 14.8mm (1.5 ratio) so for example 1000 x 667 px. Then just enter the real FL of your lens.
Will there be a Maxwell Render 6 ?

Let's be realistic. What's left of NL is only milk[…]