HOME | DD
rawimage β SCANNED AGAIN I
Published: 2003-02-06 22:11:18 +0000 UTC; Views: 3328; Favourites: 18; Downloads: 111
Redirect to original
Description
Numero uno! The time-slice-effect captured in a picture. I had a lot of fun doing those pictures. Hm, I think I should continue with this series.I used a HP ScanJet 5200C in 75 DPI prescan-mode. Then slowly rotated my head synchronously with the scan unit. added tint, a frame and type with ps 7.
*I wonder if there's a way to do this with a video camera or in post production ?? Does anybody know something?
Related content
Comments: 63
what can I say? That would be awesome! do you think it would work in realtime? the scanner took about 8 s. for such an image...
Do you have an idea at what time you'll be done with your thesis? Do you have any interest to colab with me. I would like to share my ideas of how this routine should work...
π: 0 β©: 1
hmm, well, you could capture the frames in real time, but judging from the time the automatic algorithm needed to merge just two images, the postprocessing will be slow. (as slow as photomerge of PS) But still you could use many heuristics from the known scene, i.e. track known features, use markers etc, and then I guess you could reach real time speed.
Now that I think of it if the camera is made to move on a known trajectory, much simpler methods can be used... to generate a face texture map for 3D models for instance..
My thesis is scheduled to end at the end of this summer. After that I'll be free as a bird.. well, PhD bird, not that free, but still, I like working on such practical applications and I'd especially like to hear your ideas.
π: 0 β©: 1
Are you able to check out how a twain interface or scanner driver works? I think it would be pretty much the same thing: capturing x vertical lines in a given amount of time and write them to the ram.
would your code be based on a sequence of single frames or on live video?
π: 0 β©: 1
hmm, driver coding or low level hardware manipulation is not exactly my area, except for analog cameras and webcams. Even then, the lowest level you get is capturing the whole frame, not just lines, so you have to process it, then again that's what computer vision is all about.
well, both ways are possible, it's really not much different from the coding point of view, since you essentially capture frames and work on them. sequence of frames is easier to handle, and doesn't have the real time speed constraint. you don't need a sequence of bitmaps actually. you could capture an avi and the code could break it up to bitmaps and work on them, i.e. video editing.
these all wouldn't take more than a few days in matlab, but the code could take ages to process.. in C++ one should use directX or Vfw drivers, and I'll have good templates at the end of my thesis (hopefully), plus to turn this into a tool, it has to be coded in something else than matlab, where C++ is the natural choice for me..
π: 0 β©: 1
hi lithia,
it's been a while since we started this conversation. You said you would have time when all the hustle with your exams are over. Now summer's almost over and I wanted to ask you if you're still interessted in programming a little tool to make peripheral photography. Are you?
π: 0 β©: 1
hi, i remember the conversation. my thesis is not over but i have extra time now, and i want to put my computer vision knowledge to use in this area.
so, what do you have in mind exactly? you can send me a note, that would make more sense
π: 0 β©: 0
thankyou >>> dankeschΓΆn >>> muchas gracias >>> merci >>> eeeh... mille grazie >>> huong tong chi (or whatever thanks means in chinese)
and thanks for the
π: 0 β©: 0
i have no idea, but damn thats cool. excellent job!
π: 0 β©: 1
hehe thanks a lot!
Be honest, that's definetely the latest reply you've ever received, isn't it??
π: 0 β©: 0