Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • Jul 10 13:10
    ziemek99 opened #22
  • Jul 10 11:26
    ziemek99 closed #530
  • Jul 10 11:26
    ziemek99 commented #530
  • Jun 05 15:28
    ziemek99 labeled #468
  • Jun 05 15:22
    ziemek99 labeled #481
  • Jun 04 14:41
    ziemek99 closed #145
  • Jun 04 14:41
    ziemek99 closed #150
  • Jun 04 14:41
    ziemek99 closed #278
  • Jun 04 14:40
    ziemek99 closed #420
  • Jun 04 14:40
    ziemek99 closed #209
  • Jun 04 14:40
    ziemek99 closed #350
  • Jun 04 14:40
    ziemek99 closed #425
  • Jun 04 14:39
    ziemek99 closed #475
  • Mar 09 09:59
    Maniues opened #14
  • Jan 20 18:37
    ziemek99 commented #425
  • Jan 20 18:29
    saschanaz commented #425
  • Jan 20 18:21
    ziemek99 closed #425
  • Jan 20 18:21
    ziemek99 commented #425
  • Jan 20 18:21
    ziemek99 labeled #425
  • Jan 20 18:16
    ziemek99 reopened #504
matrixbot
@matrixbot
ehmry \
Kamil Monicz
@Zaczero
@TheJaredWilcurt FLIF-hub/FLIF#499
The Jared Wilcurt
@TheJaredWilcurt
@Zaczero Nice, don't forget to update the README too, otherwise how would we follow the instructions if there are none ;)
Kamil Monicz
@Zaczero
@TheJaredWilcurt It works the same as before. The only difference is that now you can use either VS2017 or VS2015 :)
The Jared Wilcurt
@TheJaredWilcurt
:+1:
Michael Lyle
@mlyle
So, I'm doing an autonomous robot thing that's gonna communicate by satcom. I need some kind of good bit-peeled image solution. FLIF looks interesting
I've watched the couple of image races on youtube, but one is an artificial picture and the other doesn't say the absolute amount of information transferred
i need to do reasonably with a ~500 byte preview of an outside scene to know if it's worth retrieving, and then in 4k-6k get a decent picture
was wondering if anyone has any reasonable suggestions (i'm also looking at whether peeling/truncating the image makes any sense, or if the 10% or so savings over just sending the full size separately is outweighed by encoding for the specific sizes)
Christian Schneider
@schnaader
Sounds like you want to transfer images between robots instead of evaluating/processing on the first and only transferring the result (which might be below 500 bytes). So, first question: More details / are you sure transferring reduced image information is the right thing?
Michael Lyle
@mlyle
So I'm building an autonomous sailboat. Data is $1/kilobyte in the middle of the ocean
that's 1000 bytes
I want to be able to get 2-3 thumbnails back, evaluate whether the camera even worked, see if the image looks interesting :P before deciding to pay $8 for the image
Christian Schneider
@schnaader
https://data.vision.ee.ethz.ch/aeirikur/extremecompression/ looks interesting for very low bitrates, but 1) it's not completely open-sources yet and 2) it encodes "a tree" with some additional information, but for computer vision and image processing, misinformation would be too dominating
Michael Lyle
@mlyle
yah it's not for machine vision
And you know probably before even sending thumbnails there'll be some set of sanity checks on the image
Christian Schneider
@schnaader
So, thumbnails for humans. In this case, the link I posted would be best, especially since you are in a special domain (you will get non jungle images from a sailboat).
Michael Lyle
@mlyle
yah, that looks really good, thanks, and i like the comparisons
Christian Schneider
@schnaader
On the other hand, FLIF is ready for use and allows those maps where you mark "regions of quality/interest" (saliency maps), so give it a try
Michael Lyle
@mlyle
i expect most of the pictures will be boring and i won't want them, though having a wide angle thing and capturing pictures of the boat itself has value
Christian Schneider
@schnaader
Also, I'm not 100% sure about FLIF overhead, but 500 bytes FLIF will most likely transfer more information than 500 bytes JPG (which will be no information at all at that size)
https://stackoverflow.com/questions/891643/twitter-image-encoding-challenge might be worth a look, too
Michael Lyle
@mlyle
Hmm.. the extreme compression "makes up" too much of the image to be useful IMO, on further reading/evaluating
thanks for the links, though, it's good food for thought. I've got a long time fortunately to figure this out-- will be sailing a prototype at a lake this weekend
matrixbot
@matrixbot
TheGreatNi Is XMP support bugged? I tried converting to and then from flif and the metadata disappeared
matrixbot
@matrixbot
hrjet I had added XMP support to FLIF last year: FLIF-hub/FLIF#375
But the changelog for the release after that doesn't mention it: https://github.com/FLIF-hub/FLIF/releases
hrjet TheGreatNi: Which version are you using?
hrjet Also, check the version of libpng being used. IIRC, versions below 1.6 don't have the requisite support.
matrixbot
@matrixbot
TheGreatNi 0.3
TheGreatNi I'm testing it out by adding xmp data via xnview, then converting to and from flif
matrixbot
@matrixbot
hrjet make test in src/ folder has a test to verify metadata is being properly emitted. Does it pass for you?
matrixbot
@matrixbot
TheGreatNi hrjet: I'll test it out in a bit, thanks for your patience
matrixbot
@matrixbot
TheGreatNi Looks like it's the libpng version I compiled with that's the culprit, curse the repo manager
TheGreatNi Repo manager of my distro that is
Zoey Riordan
@ZoeyR
Is there a reasonable limit to put on the size of a maniac tree? Certain malicious inputs could cause endless memory growth so I want a way to stop decoding if that happens.
Zoey Riordan
@ZoeyR
Also, do maniac trees tend towards being balanced? I'm using a vector to store the nodes and that will waste a lot of space if it isn't.
Zoey Riordan
@ZoeyR
nevermind on the last question. I started storing child indicies so that the tree can be compressed to the minimum possible size.
Simanto Rahman
@Simanto_Rahman_twitter
Are there packages for linux-x64/win-x64?
Simanto Rahman
@Simanto_Rahman_twitter
Is the source code available? I needed to use FLIF with C#. Might be able to convert it into a pure C# library with managed classes if I could see the source code
The Jared Wilcurt
@TheJaredWilcurt
All of FLIF source code is available in the main repo: https://github.com/FLIF-hub/FLIF
There is a built Windows 32-bit version stored in the node-flif repo. https://github.com/FLIF-hub/node-flif
The Jared Wilcurt
@TheJaredWilcurt
@Simanto_Rahman_twitter
The Jared Wilcurt
@TheJaredWilcurt
you can see the related projects list at the bottom of the README
There are also spec documents, https://github.com/FLIF-hub/FLIF-doc and http://flif.info/papers/FLIF_ICIP16.pdf
Leo Izen
@thebombzen
has any progress been made on the whole "appending to the decoder buffer" thing?
currently if you're reading FLIF from a slow device or network stream, we can't actually reach the end of the stream, decode the truncated FLIF, then keep going once more data is available
meenu30yadav
@meenu30yadav
Hi
I want to show a demo of converting any image to flif format
Can anyone please guide me here?
The Jared Wilcurt
@TheJaredWilcurt
@meenu30yadav The official flif.exe converter only handles encoding .png, .pnm, .ppm, .pgm, .pbm, and .pamto .flif. Or decoding from .flif to one of those formats, or transcoding from .flif to .flif