421
edits
No edit summary |
|||
(15 intermediate revisions by the same user not shown) | |||
Line 503: | Line 503: | ||
==|'''WEEK EIGHT'''| & |'''WEEK NINE'''|== | ==|'''WEEK EIGHT'''| & |'''WEEK NINE'''|== | ||
'''Slime mold''' | |||
I left Weimar for Christmas break. But I want to show some slime mold pictures under the microscope that I personally took in the Biolab after some days of them growing. I think they are pretty amazing. | |||
<gallery> | |||
File:slimemold1.jpg | |||
File:slimemold2.jpg | |||
File:slimemold3.jpg | |||
File:slimemold4.jpg | |||
File:slimemold6.jpg | |||
File:slimemold7.jpg | |||
File:slimemold8.jpg | |||
File:slimemold9.jpg | |||
File:slimemold10.jpg | |||
File:slimemold11.jpg | |||
File:slimemold12.jpg | |||
File:slimemold13.jpg | |||
File:slimemold14.jpg | |||
File:sls.jpg | |||
File:slimemold15.jpg | |||
File:slimemold16.jpg | |||
File:slimemold17.jpg | |||
File:slimemold18.jpg | |||
File:slimemold19.jpg | |||
File:slimemold20.jpg | |||
File:slimemold21.jpg | |||
File:slimemold22.jpg | |||
File:slimemold23.jpg | |||
File:slimemold24.jpg | |||
File:slimemold25.jpg | |||
File:slimemold27.jpg | |||
File:slimemold28.jpg | |||
File:slimemold29.jpg | |||
File:slimemold30.jpg | |||
File:slimemold31.jpg | |||
File:slimemold32.jpg | |||
File:slimemold33.jpg | |||
File:slimemold34.jpg | |||
File:slimemold35.jpg | |||
File:slimemold36.jpg | |||
File:slimemold37.jpg | |||
File:slimemold38.jpg | |||
File:slimemold39.jpg | |||
File:slimemold40.jpg | |||
File:slimemold41.jpg | |||
File:slimemold42.jpg | |||
File:slimemold43.jpg | |||
File:slimemold44.jpg | |||
File:slimemold45.jpg | |||
File:slimemold46.jpg | |||
File:slimemold47.jpg | |||
File:slimemold48.jpg | |||
File:slimemold49.jpg | |||
</gallery> | |||
==='''AI generated images'''=== | |||
I collected many microscopy pictures so far. I have a dataset of more than 2300 images. I decided to lend a new slant to this project, moving from analog to digital. | I collected many microscopy pictures so far. I have a dataset of more than 2300 images. I decided to lend a new slant to this project, moving from analog to digital. | ||
I worked with an online software which takes advantage of machine learning (to be more precise StyleGan2) to synthetise new data starting from a massive dataset. Hence, I trained several different machine learning models started from different pre-trained dataset to start learning from, and then added my original microscopy dataset. | I worked with an online software which takes advantage of machine learning (to be more precise StyleGan2) to synthetise new data starting from a massive dataset. Hence, I trained several different machine learning models started from different pre-trained dataset to start learning from, and then added my original microscopy dataset. | ||
I selected a parameter called '''steps''' which influences in a significant way the training process: it establishes the number of steps the model has to go through, so the timing required (of course the more it works the better.) | I selected a parameter called '''steps''' which influences in a significant way the training process: it establishes the number of steps the model has to go through, so the timing required (of course the more it works the better.) | ||
---- | |||
--- | |||
'''1st attempt''' 3000 steps | '''1st attempt''' 3000 steps | ||
--- | |||
[[ | [[File:steps-1st-attemp.mp4]] | ||
Here I started from a pre-trained dataset of HQ landscapes and then selected my first microscopy dataset as custom dataset. The first dataset was based on less images - about 1200 - therefore I got as a result a less heterogeneous folder of images. | Here I started from a pre-trained dataset of HQ landscapes and then selected my first microscopy dataset as custom dataset. The first dataset was based on less images - about 1200 - therefore I got as a result a less heterogeneous folder of images. | ||
Line 518: | Line 580: | ||
1000 new samples images generated by this machine learning model which look like my images (although not that similar because based on an indefinite shape; this feature comes from the fact that my original images where quite blurred and dark, in fact my initial aim was to create analog errors! It is pretty amazing how the algorithm works in trying to reproduce the input). | 1000 new samples images generated by this machine learning model which look like my images (although not that similar because based on an indefinite shape; this feature comes from the fact that my original images where quite blurred and dark, in fact my initial aim was to create analog errors! It is pretty amazing how the algorithm works in trying to reproduce the input). | ||
Preview generated folder: | |||
<gallery> | |||
File:img000000002.jpg | |||
File:img000000015.jpg | |||
File:img000000034.jpg | |||
File:img000000058.jpg | |||
File:img000000064.jpg | |||
File:img000000065.jpg | |||
File:img000000128.jpg | |||
File:img000000171.jpg | |||
File:img000000146.jpg | |||
File:img000000529.jpg | |||
File:img000000723.jpg | |||
File:img000000816.jpg | |||
</gallery> | |||
---- | |||
--- | |||
'''2nd attempt''' 7000 steps | '''2nd attempt''' 7000 steps | ||
--- | |||
[[File:steps-dataset2-microscopy-2ndattempt - Progress Video.mp4]] | |||
Here I started from a pre-trained dataset of HQ landscapes and then selected my second microscopy dataset as custom one. The second dataset was based on more images - 2300 - therefore I got as a result a more heterogeneous folder of images. | Here I started from a pre-trained dataset of HQ landscapes and then selected my second microscopy dataset as custom one. The second dataset was based on more images - 2300 - therefore I got as a result a more heterogeneous folder of images. | ||
Line 527: | Line 611: | ||
1000 new samples images generated by this machine learning model which look like my images (they still have indefinite shapes instead of rounded ones, but they are more colorful and diverse); | 1000 new samples images generated by this machine learning model which look like my images (they still have indefinite shapes instead of rounded ones, but they are more colorful and diverse); | ||
Preview generated folder: | |||
<gallery> | |||
File:img000000000.jpg | |||
File:img000000006.jpg | |||
File:img000000003.jpg | |||
File:img000000005.jpg | |||
File:img000000030.jpg | |||
File:img000000018.jpg | |||
File:img000000044.jpg | |||
File:img000000088.jpg | |||
File:img000000095.jpg | |||
File:img000000001.jpg | |||
File:img000000205.jpg | |||
File:img000000180.jpg | |||
</gallery> | |||
---- | |||
--- | |||
'''3rd attempt''' 700 steps | '''3rd attempt''' 700 steps | ||
--- | |||
[[File:dataset-3-microscopy-dn - Progress Video.mp4]] | [[File:dataset-3-microscopy-dn - Progress Video.mp4]] | ||
Line 537: | Line 643: | ||
500 new samples images generated by this machine learning model which look like my images (they still have indefinite shapes instead of rounded ones, but they are more colorful and diverse); | 500 new samples images generated by this machine learning model which look like my images (they still have indefinite shapes instead of rounded ones, but they are more colorful and diverse); | ||
Preview generated folder: | |||
<gallery> | |||
File:gifedwd.jpg | |||
File:gifedwd1.jpg | |||
File:gifedwd2.jpg | |||
File:gifedwd3.jpg | |||
File:gifedwd4.jpg | |||
File:gifedwd5.jpg | |||
File:gifedwd6.jpg | |||
File:gifedwd7.jpg | |||
File:gifedwd16.jpg | |||
File:gifedwd62.jpg | |||
File:gifedwd99.jpg | |||
File:v25.jpg | |||
</gallery> | |||
---- | |||
--- | |||
'''4th attempt''' 8000 steps: | '''4th attempt''' 8000 steps: | ||
--- | |||
[[File:Training Image Experiment on mycroscope - Progress Video.mp4]] | |||
Here I decided to change the pre-trained dataset with closer data in order to see whether I could get an improvement. Therefore, I used the 1st folder generated by the ML model as pre-trained dataset and then selected the complete microscope dataset as custom one. | Here I decided to change the pre-trained dataset with closer data in order to see whether I could get an improvement. Therefore, I used the 1st folder generated by the ML model as pre-trained dataset and then selected the complete microscope dataset as custom one. | ||
Line 546: | Line 674: | ||
1000 new samples images generated by this machine learning model which are really similar to my original images, rounded and detailed. | 1000 new samples images generated by this machine learning model which are really similar to my original images, rounded and detailed. | ||
--- | Preview generated folder: | ||
<gallery> | |||
File:gdeiugcf.jpg | |||
File:hbwhjdbw08.jpg | |||
File:hbwhjdbw12.jpg | |||
File:hbwhjdbw26.jpg | |||
File:hbwhjdbw44.jpg | |||
File:hbwhjdbw49.jpg | |||
File:hbwhjdbw50.jpg | |||
File:hbwhjdbw67.jpg | |||
File:hbwhjdbw82.jpg | |||
File:hbwhjdbw134.jpg | |||
File:hbwhjdbw187.jpg | |||
File:hbwhjdbw721.jpg | |||
</gallery> | |||
---- | |||
Now I am trying to create some videos by shaping interpolation on the latent space, so smooth transitions between one generated picture and another, with different soundtracks under them. | Now I am trying to create some videos by shaping interpolation on the latent space, so smooth transitions between one generated picture and another, with different soundtracks under them. | ||
Line 553: | Line 699: | ||
''07/01/2021'' | ''07/01/2021'' | ||
Some of the interpolation videos I made with different soundtracks over: | |||
[https://cloud.uni-weimar.de/s/46XRnyKLwBsG9cf?path=%2Fdenise%20nicolau%2FML-videos Video1-bauhaus] | |||
[https://cloud.uni-weimar.de/s/46XRnyKLwBsG9cf?path=%2Fdenise%20nicolau%2FML-videos Video2-teebs] | |||
[https://cloud.uni-weimar.de/s/46XRnyKLwBsG9cf?path=%2Fdenise%20nicolau%2FML-videos Video3-nude] | |||
[https://cloud.uni-weimar.de/s/46XRnyKLwBsG9cf?path=%2Fdenise%20nicolau%2FML-videos Video4-robotics] | |||
[https://cloud.uni-weimar.de/s/46XRnyKLwBsG9cf?path=%2Fdenise%20nicolau%2FML-videos Video5-robotics] | |||
[https://cloud.uni-weimar.de/s/46XRnyKLwBsG9cf?path=%2Fdenise%20nicolau%2FML-videos Video6-sadwhales] | |||
==|'''WEEK ELEVEN'''|== | |||
''12/01/2021'' | |||
I made two others interpolation videos putting as a soundtrack a BBC documentary: | |||
1st: [https://cloud.uni-weimar.de/s/46XRnyKLwBsG9cf?path=%2Fdenise%20nicolau%2FML-videos Blue Planet] | |||
2nd: [https://cloud.uni-weimar.de/s/46XRnyKLwBsG9cf?path=%2Fdenise%20nicolau%2FML-videos Black Holes] | |||
''15/01/2021'' | |||
My collection of videos is done. I thought it was nice to create a loop video with some of the interpolations playing together all in once, and so I did > [https://cloud.uni-weimar.de/s/46XRnyKLwBsG9cf?path=%2Fdenise%20nicolau%2FML-videos Loop Interpolation Videos] | |||
==|'''WEEK TWELVE'''|== | |||
''18/01/2021'' | |||
I edited two videos in speed and sound, adjusting some parameters to better match visual and sound: | |||
[https://cloud.uni-weimar.de/s/46XRnyKLwBsG9cf?path=%2Fdenise%20nicolau%2FML-videos black holes] | |||
[https://cloud.uni-weimar.de/s/46XRnyKLwBsG9cf?path=%2Fdenise%20nicolau%2FML-videos blue planet 2] | |||
''20/01/2021'' | |||
I am creating a narrative that connects each video avoiding the linear common sense, in favor of dizzying the spectator. \WORK IN PROGRESS\ | |||
==|'''WEEK THIRTEEN & FOURTEEN'''|== | |||
''03/02/2021'' | |||
The final narrative is ready here: [https://www.youtube.com/watch?v=awOCMRase8s&ab_channel=DeniseNicolau Echoes] |
edits