samim23 / deepdreamanim Goto Github PK
View Code? Open in Web Editor NEWDeepDream Animation Helper
License: MIT License
DeepDream Animation Helper
License: MIT License
Noob with little experience here,
When trying to run any comand i get this same error
File "E:\AITHINGS\DeepDreamAnim-master\dreamer.py", line 80
print octave, i, end, vis.shape
^
SyntaxError: Missing parentheses in call to 'print'. Did you mean print(octave, i, end, vis.shape)?
Any help?
Thanks!
Please help, I'd love to try out your software. I am currently running a conda environment, and was able to install all dependencies except for caffe.
Hi, thanks for the excellent script - I've got it working fine except that when I try to set a guide image this error is thrown:
File "dreamer.py", line 21, in preprocess
return np.float32(np.rollaxis(img, 2)[::-1]) - net.transformer.mean['data']
ValueError: operands could not be broadcast together with shapes (4,301,600) (3,1,1)
I've tried a variety of .png images (even trying one of the source images) but I get this error every time.
I am a bit new to this, but when I ran it, it would run through the inception layers just fine how it normally would, towards the last few lines I got the following error.
Traceback (most recent call last):
File "dreamer.py", line 254, in
main(args.input, args.output, args.preview, args.octaves, args.octavescale, args.iterations, args.jitter, args.zoom,
args.stepsize, args.blend, args.layers)
File "dreamer.py", line 169, in main
img = PIL.Image.open(inputdir+'/'+vids[0])
IndexError: list index out of range
First up, big fan of your work, and I'm very excited about getting this working.
When I try to actually process the frames (I believe the command line should look something like) "C:\Anaconda>python dreamer.py --input frames\bp --output pic" it goes through some parts of the operation and then fails.
Traceback (most recent call last):
File "dreamer.py", line 339, in
args.zoom, args.stepsize, args.blend, args.layers, args.guide, args.gpu, args.flow)
File "dreamer.py", line 166, in main
img = PIL.Image.open(inputdir + '/' + vids[0])
IndexError: list index out of range
As I understand it, "indexerror: out of range" is that it usually is in reference to a line that isn't there, but line 339 is totally there, and that appears to be the bug reported in #3 , and the closing comment says that the index error is an image processing the last frame. Which is odd, because it isn't processing any frames at the moment.
The research I've done about the pil error has not turned up an answer that I've been able to put to solving the problem.
I've been talking to a friend who has had success with running deepdreamanim, he said that it could be related to my caffe paths, but as far as I can tell, he and I are using the same pathing.
Lines 134-138
model_name = 'bvlc_googlenet'
model_path = 'C:/anaconda/caffe/models/' + model_name + '/'
net_fn = model_path + 'deploy.prototxt'
param_fn = model_path + 'bvlc_googlenet.caffemodel'
I've also tried adding a 0 before each file name, but have not tried the files without the zero.
Any and all help is appreciated. Thanks.
When flow is enabled, specially for sequences where frames are similar (slow or no camera movement, partial/slow pixel movement) every frame becomes additive. Is this the normal behavior ?
When flow is turned off, everything seems okay over a long period of time (but there is a lot of jitter for sure) but when flow is on it becomes brighter and brighter every frame unless the changes between frames are too clear.
Hi. After some installation experimentation, I've managed to get all the dependencies installed under Windows 7: Anaconda, FFmpeg, CUDA, CudNN, Caffe, + all the addons etc required. I've accomplished this following tutorials which are now linked on new wiki page in the DeepDreamAnim wiki.
So far so good. I stuck the dreamer.py file in the same parent directory as caffe and have managed to get through clipping up the volcano using:
C:\deepdream\DreamAnim>python dreamer.py -i C:\frames -o C:\frames_out
and then successfully dreampt onto those frames after swapping the places of the frames to new folders with:
C:\deepdream\DreamAnim>python dreamer.py -i C:\frames -o C:\frames_out
However, the last step generates an error message:
[image2 @ 0000000002f9f5c0] Pattern type 'glob' was selected but globbing is not supported by this libavformat build
C:\frames_out/*.jpeg: Function not implemented
This really won't be an issue in terms of generating videos for me, as I'm quite comfortable using myfavoritesoftware to do the stitching (like Blender).
Is this something in the dreamer.py or simply something to do with running this under Windows?
Warm Regards!
M
P.S. Great Job!
Currently dreaming a video with 1188 frames and I'm noticing weird things once I passed 100 frames.
Everything is in order with the input frames but the dreamed frames…
10099__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime9s_filenameimage-099.jpeg
10100__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime9s_filenameimage-100.jpeg
10101__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime9s_filenameimage-1000.jpeg
10102__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime9s_filenameimage-1001.jpeg
10103__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime9s_filenameimage-1002.jpeg
10104__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime9s_filenameimage-1003.jpeg
10105__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime9s_filenameimage-1004.jpeg
10106__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime9s_filenameimage-1005.jpeg
10107__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime9s_filenameimage-1006.jpeg
10108__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime9s_filenameimage-1007.jpeg
10109__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime9s_filenameimage-1008.jpeg
10110__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime9s_filenameimage-1009.jpeg
10111__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime9s_filenameimage-101.jpeg
10112__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime9s_filenameimage-1010.jpeg
10113__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime9s_filenameimage-1011.jpeg
10114__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime9s_filenameimage-1012.jpeg
10115__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime9s_filenameimage-1013.jpeg
10116__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime9s_filenameimage-1014.jpeg
10117__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime9s_filenameimage-1015.jpeg
10118__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime9s_filenameimage-1016.jpeg
10119__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime9s_filenameimage-1017.jpeg
10120__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime9s_filenameimage-1018.jpeg
10121__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime11s_filenameimage-1019.jpeg
10122__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime10s_filenameimage-102.jpeg
10123__octaves4_iterations10_octavescale1.5_netinception_4c-output_jitter32_stepsize1.5_blend0.5_renderTime11s_filenameimage-1020.jpeg
Every tenth frame after 100 ends in 3 numbers instead of 4 and those frames contain a frame from earlier in the sequence, blended in correctly but definitely misplaced.
I tested your setup and it is all working nicely, GPU works too which is cool.
The problem is that, the dream just stops at some point. Only after few frames, and what left is the dream for previous frames as ghost image for the following frames.
https://youtu.be/xtkG_lftImg
Any idea why this happens? Could it be because of screensaver?
I keep getting this error when trying to use it
(caffe) C:\Users\root\DeepDreamAnim>python dreamer.py --input myvideo --output myvideo/frames --preview 600
Traceback (most recent call last):
File "dreamer.py", line 339, in <module>
args.zoom, args.stepsize, args.blend, args.layers, args.guide, args.gpu, args.flow)
File "dreamer.py", line 143, in main
text_format.Merge(open(net_fn).read(), model)
IOError: [Errno 2] No such file or directory: '../../caffe/models/bvlc_googlenet/deploy.prototxt'
having trouble using the guide command and I always get this outcome:
Traceback (most recent call last):
File "dreamer.py", line 339, in <module>
args.zoom, args.stepsize, args.blend, args.layers, args.guide, args.gpu, args.flow)
File "dreamer.py", line 173, in main
guideimg = PIL.Image.open(inputdir + '/' + guide)
File "C:\Users\root\Anaconda2\envs\caffe\lib\site-packages\PIL\Image.py", line 1952, in open
fp = __builtin__.open(fp, "rb")
IOError: [Errno 22] invalid mode ('rb') or filename: 'C:\\Users\\root\\Desktop\\input/C:\\Users\\root\\Desktop\\input\\image-000406.png'
TypeError: Required argument 'flow' (pos 3) not found
I have cv2 installed -- any ideas?
When I run the image processing pass, the script seems to pull random frames from the frame directory instead of processing them sequentially:
Frame 158 of 1866
Frame Time: 1s
Estimated Total Time Remaining: 1708s (0:28:28)
Processing: myvideo/image-000659.png
0 0 inception_3b/output (360, 540, 3)
0 1 inception_3b/output (360, 540, 3)
...
3 9 inception_3b/output (360, 540, 3)
Saving Image As: myvideo/frames/frame_000159.png
this gives me an unwatchable mess when I try to re-stitch the frames.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.