
sorry i wasn't sure which part you meant. I tend to over explain...please don't take it as me thinking you're unintelligent..it's just when i read instructions i prefer to have over explained then under-explain if that's a word.
the automation part?
I downloaded a tool called GhostMouse. When that tool runs, it has a record button, a play button, and a stop button.
I hit record then manually do the steps describe switching between gimp and the web browser.
instead of typing in filenames for example 001_input.jpg and dream_001.jpg, I created a spreadsheet on google(before recording of course).
on the spread sheet.
- for the first column i start with 1 and 2 as the first two cells then drag the bottom right corner or column 1, row 2 down to how ever many frames/rows i want and it populates all numbers upward to that number of rows (for example down to row 200).
- then in the second column, I put =concat("000",A1) in the first row, then drag the bottom right corner of that cell down to row 200.
- then in the third column, I put =concat(RIGHT(B1,3),"_input.jpg"), then drag the bottom right corner of that cell down to row 200 (now i have all my input file names generated in 3rd column)
- then in the fourth column, I put =concat(concat("dream_",right(B1,3)),".jpg"), then drag the bottom right corner of that cell down to 200 (now i have all my output file names generated in 4th column)
- I then insert a blank row on top, and select the 4 column first empty row to pretend like i am on the output file this way when i start recording my actions i can just move between cells using arrow keys and work downward.
Then i click the record button in GhostMouse.
And then i do what i need to do that i want GhostMouse to repeat next time.
like opening up an image in gimp (I already tried opening the image before recording in the folder that i want output files to save to before recording) and made sure that the names were decending in order so that if i pick the first file it would always select the lastest dream_00X.jpg file.
i then layer scale blah blah blah ...
then before exporting the file, i would switch to my spreadsheet (in the browser) and move to the next row in column 3 (using arrow keys not mouse), copy that filename to use in the exporting of gimp so i don't have to type in a filename i just paste a value in there.
then close the file in gimp so that next time i repeat my actions everything would be in the same starting point.
switch to browser with deepdream already open and already generated an image (because i want it to be repeatable) click on top left corner to get the starting page, waited 3 seconds, click on button on page and pasted the same filename already copied as exported file and then waited 15 seconds (maybe a little longer just to be sure in case internet is slow sometimes) go to my browser tab with spreadsheet, move to column4(using keyboard) copy output file name, go back to deepdream result page, and save as then pasted the copied filename in there.
and then click stop button in GhostMouse
I then close down GhostMouse, it asks to save, I click YES and give it a filename.rms and saved to my desktop.
I then edit the .rms file using a text editor
I deleted the last 3 lines that says something like
{LMouse down (530,449)}
{Delay 0.09}
{LMouse up (530,449)}
since that's the action for when I clicked on the Stop button in GhostMouse (I don't want that repeated).
I then copied the whole thing, pasted it as many times as I want, in my case 200 times (I don't think there's a way to repeat how many times you want to play in GhostMouse other than pasting it multiple times at least i don't know how)
I then save the .rms close it.
and now double click the .rms file it automatically opens up GhostMouse.
I then click play and watch it go to work for a couple of frames making sure it works then leave the computer for hours and come back when it's done.
I would share the .rms file if it made sense but because each computer is different and i have different things opened in different locations on the screen, the .rms won't be of use to anyone but on the same screen and apps opened as when it was recorded.
Hope that explained enough to get you going if you wanted to do something similar... If I was doing another one I'd definitely let it run overnight.... 15 seconds here and 15 seconds there adds up so it takes the same amount of time as if you were sitting there doing it manually yourself but it works... I am sure there's an easier way if that deepdream page had like an API or something and you could program it it'd probably take much less time.