This is a Python program to produce images or videos.
It extracts random (or sequential) frames from a video or image.
It (optionally) places words somewhere on each frame.
Then joins all frames into an animation or image.
You can use many arguments to produce different kinds of results.
It might be useful in the realm of human verification.
And memes.
pipx install git+this_repo_url --forceNow you should have the gifmaker command available.
Clone this repo, and get inside the directory:
git clone --depth 1 this_repo_url
cd gifmakerThen create the virtual env:
python -m venv venvThen install the dependencies:
venv/bin/pip install -r requirements.txtOr simply run scripts/venv.sh to create the virtual env and install the dependencies.
There's a scripts/test.sh file that runs the program with some arguments to test if things are working properly.
Use the installed gifmaker command if you used pipx.
Or run gifmaker/main.py using the Python in the virtual env:
venv/bin/python -m gifmaker.mainThere's a run.sh that does this.
You can provide a video or image path using the --input argument.
You also need to provide an output path:
gifmaker --input /path/to/video.webm --output /tmp/gifmaker
gifmaker --input /path/to/animated.gif --output /tmp/gifmaker
gifmaker --input /path/to/image.png --output /tmp/gifmakerwebm, mp4, gif, jpg, and png should work, and maybe other formats.
You can pass it a string of lines to use on each frame.
They are separated by ; (semicolons).
gifmaker --words "Hello Brother ; Construct Additional Pylons"It will make 2 frames, one per line.
If you want to use words and have some frames without them simply use more ;.
You can use random words with [random]:
gifmaker --words "I Like [random] and [random]"It will pick random words from a list of English words.
There are 4 kinds of random formats: [random], [RANDOM], [Random], [RanDom], and [randomx].
The replaced word will use the case of those.
With [random] you get lower case, like water.
With [RANDOM] you get all caps, like PLANET.
With [Random] you get the first letter capitalized, like The garden.
With [RanDom] you get title case, like The Machine.
With [randomx] you get the exact item from the random list.
You can specify how many random words to generate by using a number:
For example [Random 3] might generate Drivers Say Stories.
You can multiply random commands by using numbers like [x2].
For example:
--words "Buy [Random] [x2]"This might produce: Buy Sink ; Buy Plane.
The multipliers need to be at the end of the line.
You can also generate random numbers with [number].
This is a single digit from 0 to 9.
For example, [number] might result in 3.
You can specify the length of the number.
For example, [number 3] might result in 128.
You can also use a number range by using two numbers.
For example, [number 0 10] will pick a random number from 0 to 10.
--words "I rate it [number 0 10] out of 10"If you want to repeat the previous line, use [repeat]:
For example: --words "Buy Buttcoin ; [repeat]".
It will use that text in the first two frames.
You can also provide a number to specify how many times to repeat:
For example: --words "Buy Buttcoin ; [repeat 2]".
The line will be shown in 3 frames (the original plus the 2 repeats).
A shortcut is also available: [rep] or [rep 3].
You can use linebreaks with \n.
For example: --words "Hello \n World".
Will place Hello where a normal line would be.
And then place World underneath it.
Another way to define an empty line is using [empty].
For example: hello ; world ; [empty].
This could be useful in wordfile to add empty lines at the end.
Else you can just add more ; to words.
You can also use numbers like [empty 3].
That would add 3 empty frames.
There's also [date] which can be used to print dates.
You can define any date format in it.
For example [date %Y-%m-%d] would print year-month-day.
You can see all format codes here: datetime docs.
If no format is used it defaults to %H:%M:%S.
There's also [count].
The count starts at 0 and is increased on every [count].
For example --words "Match: [count] ; Nothing ; Match [count]".
It would print Match: 1, Nothing, and Match: 2.
You might want to print the count on every frame:
--words "[count]" --fillgen --frames 10You can run main.py from anywhere in your system using its virtual env.
Relative paths should work fine.
If you provide an argument without flags, it will be used for words:
gifmaker --input image.png "What is that? ; AAAAA?"It's a shortcut to avoid having to type --words.
Here's a fuller example:
gifmaker --input /videos/stuff.webm --fontsize 18 --delay 300 --width 600 --words "I want to eat ;; [Random] ; [repeat 2] ;" --format mp4 --bgcolor 0,0,0 --output stuff/videos
You can use arguments like: --delay 350 --width 500 --order normal.
These modify how the file is going to be generated.
input (Type: str | Default: None)
Path to a video or image to use as the source of the frames.
webm, mp4, gif, and even jpg or png should work.
For example: --input stuff/cow.webm.
-i is a shorter alias for this.
output (Type: str | Default: None)
Directory path to save the generated file.
For example: stuff/videos.
It will use a random file name.
Using gif, webm, mp4, jpg, or png depending on the format argument.
Or you can enter the path plus the file name.
For example: stuff/videos/cat.gif.
The format is deduced by the extension (gif, webm, mp4, jpg, or png).
-o is a shorter alias for this.
words (Type: str | Default: Empty)
The words string to use.
Lines are separated by ;.
Each line is a frame.
Special words include [random] and [repeat].
As described in Usage.
wordfile (Type: str | Default: None)
File to use as the source of word lines.
For example, a file can be like:
This is a line
I am a [random]
This is a line after an empty line
[repeat]
[empty]
Then you can point to it like:
--wordfile "/path/to/words.txt"It will use word lines the same as with --words.
fillwords (Type: flag | Default: False)
Fill the rest of the frames with the last word line.
If there are no more lines to use, it will re-use the last line.
For example:
--words "First Line; Last Line" --frames 5 --fillwordsFirst frame says "First Line".
Then it will use "Last Line" for the rest of the frames.
fillgen (Type: flag | Default: False)
If this is enabled, the first line of words will be generated till the end of the frames.
For example:
gifmaker --words "[random] takes [count] at [date]" --fillgen --frames 5separator (Type: str | Default: ";")
The character to use as the line separator in words.
This also affects randomlist.
delay (Type: int | Default: 700)
The delay between frames. In milliseconds.
A smaller delay = A faster animation.
frames (Type: int | Default: 3)
The amount of frames to use.
This value has a higher priority than the other frame count methods.
framelist (Type: str | Default: Empty)
The specific list of frame indices to use.
The first frame starts at 0.
For example --framelist "2,5,2,0,3".
It will use those specific frames.
It also defines how long the animation is.
frameopts (Type: str | Default: Empty)
Define the pool of frame indices when picking randomly.
For example: --frameopts 0,11,22.
left (Type: int | Default: None)
Padding from the left edge to position the text.
right (Type: int | Default: None)
Padding from the right edge to position the text.
top (Type: int | Default: None)
Padding from the top edge to position the text.
bottom (Type: int | Default: None)
Padding from the bottom edge to position the text.
You only need to set left or right, not both.
You only need to set top or bottom, not both.
If those are not set then the text is placed at the center.
If any of those is set to a negative value like -100, it will apply it from the center.
For example: --top -100 would pull it a bit to the top from the center.
And --right -100 would pull it a bit to the right from the center.
width (Type: int | Default: None)
Fixed width of every frame.
If the height is not defined it will use an automatic one.
height (Type: int | Default: None)
Fixed height of every frame.
If the width is not defined it will use an automatic one.
nogrow (Type: flag | Default: False)
If this is enabled, the frames won't be resized if they'd be bigger than the original.
For instance, if the original has a width of 500 and you set --width 600.
It's a way to limit the values of --width and --height.
format (Type: str | Default: "gif")
The format of the output file. Either gif, webm, mp4, jpg, or png.
This is only used when the output is not a direct file path.
For instance, if the output ends with cat.gif it will use gif.
If the output is a directory it will use a random name with the appropriate format.
order (Type: str | Default: "random")
The order used to extract the frames.
Either random or normal.
random picks frames randomly.
normal picks frames in order starting from the first one.
normal loops back to the first frame if needed.
font (Type: str | Default "sans")
The font to use for the text.
Either: sans, serif, mono, bold, italic, cursive, comic, or nova.
There is also random and random2.
First one is a random font, the other one is a random font on each frame.
You can also specify a path to a ttf file.
fontsize (Type: int | Default: 60)
The size of the text.
fontcolor (Type: str | Default: "255,255,255")
The color of the text.
This is a color.
bgcolor (Type: str | Default: None)
Add a background rectangle below the text.
This is a color.
opacity (Type: float | Default: 0.66)
From 0 to 1.
The opacity level of the background rectangle.
The closer it is to 0 the more transparent it is.
padding (Type: int | Default: 20)
The padding of the background rectangle.
This gives some spacing around the text.
radius (Type: int | Default: 0)
The border radius of the background rectangle.
This is to give the rectangle rounded corners.
outline (Type: str | Default: None)
Add an outline around the text.
In case you want to give the text more contrast.
This is a color.
outlinewidth (Type: int | Default: 2)
Width of the outline. It must be a number divisible by 2.
If it's not divisible by 2 it will pick the next number automatically.
no-outline-top (Type: flag | Default: False)
no-outline-bottom (Type: flag | Default: False)
no-outline-left (Type: flag | Default: False)
no-outline-right (Type: flag | Default: False)
Don't show specific lines from the outline.
align (Type: str | Default: "center")
How to align the center when there are multiple lines.
Either left, center, or right.
wrap (Type: int | Default: 35)
Split lines if they exceed this char length.
It creates new lines. Makes text bigger vertically.
nowrap (Type: flag | Default: False)
Don't wrap the lines of words.
randomlist (Type: str | Default: Empty)
Random words are selected from this list.
If the list is empty it will be filled with a long list of nouns.
You can specify the words to consider, separated by semicolons.
For example: --randomlist "cat ; dog ; nice cow ; big horse".
randomfile (Type: str | Default: List of nouns)
Path to a text file with the random words to use.
This is a simple text file with each word or phrase in its own line.
For example:
dog
a cow
horse
Then you point to it: --randomfile "/path/to/animals.txt".
repeatrandom (Type: flag | Default: False)
If this is enabled, random words can be repeated at any time.
Else it will cycle through them randomly without repetitions.
loop (Type: int | Default 0)
How to loop gif renders.
-1 = No loop
0 = Infinite loop
1 or more = Specific number of loops
filter (Type: str | Default: "none")
A color filter that is applied to each frame.
The filters are: hue1, hue2 .. up to hue8, and anyhue, anyhue2.
And also: gray, blur, invert, random, random2, none.
random picks a random filter for all frames.
random2 picks a random filter on every frame.
anyhue is like random but limited to the hue effects.
anyhue2 is like random2 but is limited to the hue effects.
filteropts (Type: str | Default: Empty)
This defines the pool of available filters to pick randomly.
This applies when filter is random or random2.
For example: --filteropts hue1,hue2,hue3,gray.
repeatfilter (Type: flag | Default: False)
If this is enabled, random filters can be repeated at any time.
Else it will cycle through them randomly without repetitions.
remake (Type: flag | Default: False)
Use this if you only want to re-render the frames.
It re-uses all the frames, resizes, and renders again.
It doesn't do the rest of the operations.
For example: --input /path/to/file.gif --remake --width 500 --delay 300.
For instance, you can use this to change the width or delay of a rendered file.
descender (Type: flag | Default: False)
If enabled, the descender height will add extra space to the bottom of text.
This is relevant when adding a background or an outline.
This means words like Ayyy get covered completely, top of A and bottom of y.
The default is to ignore the descender to ensure consistent placement of text.
seed (Type: int | Default: None)
frameseed (Type: int | Default: None)
wordseed (Type: int | Default: None)
filterseed (Type: int | Default: None)
colorseed (Type: int | Default: None)
The random component can be seeded.
This means that if you give it a value, it will always act the same.
This can be useful if you want to replicate results.
There are multiple random generators:
One takes care of picking frames and is controlled by frameseed.
One takes care of picking words and numbers and is controlled by wordseed.
One takes care of picking filters and is controlled by filterseed.
One takes care of picking colors and is controlled by colorseed.
If those are not defined, then it will assign the generic seed (if defined).
If no seed is defined then it won't use seeds and be truly random (the default).
deepfry (Type: flag | Default: False)
Apply heavy jpeg compression to all frames.
Use to distort the result for whatever reason.
word-color-mode (Type: str | Default: "normal")
Either normal or random.
In normal it will avoid fetching random colors on the same lines.
For instance if the words are First line [x2] ; Second line [x2].
And the colors are set to --fontcolor light2 --bgcolor darkfont2.
It will use the same colors for the first 2 frames, and then other colors for the rest.
Instead of picking random colors on each frame.
This is to avoid the text being too aggresive visually.
This can be disabled with random, which will fetch random colors on each frame.
If a number argument has a default you can use p and m operators.
p means plus while m means minus.
For example, since fontsize has a default of 20.
You can do --fontsize p1 or --fontsize m1.
To get 21 or 19.
Some arguments use the color format.
This can be 3 numbers from 0 to 255, separated by commas.
It uses the RGB format.
0,0,0 would be black, for instance.
The value can also be light or dark.
These will get a random light or dark color.
The value can also be light2 or dark2.
These will get a random light or dark color on each frame.
Names are also supported, like green, black, red.
It can also be font to use the same color as the font.
It can also be lightfont and darkfont.
These pick contrasts based on the current font color.
For example if the font color is light red, the contrast would be dark redish.
lightfont2 and darkfont2 do the same but on each frame.
You can make TOML files that define the arguments to use.
Provide the path of a script like this: --script "/path/to/script.toml".
For example, a script can look like this:
words = "Disregard [Random] ; [repeat] ; Acquire [Random] ; [repeat] ;"
fontcolor = "44,80,200"
fontsize = 80
bgcolor = "0,0,0"
bottom = 0
right = 0You can write shell functions to make things faster by using templates.
For example here's a fish function:
function funstuff
gifmaker \
--input /path/to/some/file.png --words "$argv is [Random] [x5]" \
--bgcolor random_dark2 --fontcolor random_light2 \
--top 0 --fontsize 22 --filter random2 --width 600
endThis is added in ~/.config/fish/config.fish.
Source the config after adding the function:
source ~/.config/fish/config.fishThen you can run: funstuff Grog.
In this case it will do Grogg is [Random] 5 times.
Using all the other arguments that are specific to look good on that image.
You might want to interface through another Python program.
Here's some snippets that might help:
import asyncio
from pathlib import Path
# Arguments shared by all functions
gifmaker_common = [
"/usr/bin/gifmaker",
"--width", 350,
"--output", "/tmp/gifmaker",
"--nogrow",
]
# Add quotes around everything and join
def join_command(command):
return " ".join(f'"{arg}"' for arg in command)
# Get the command list and turn it into a string
def gifmaker_command(args):
command = gifmaker_common.copy()
command.extend(args)
return join_command(command)
# You can have multiple functions like this
def generate_something(who):
command = gifmaker_command([
"--input", get_path("describe.jpg"),
"--words", f"{who} is\\n[Random] [x5]",
"--filter", "anyhue2",
"--opacity", 0.8,
"--fontsize", 66,
"--delay", 700,
"--padding", 50,
"--fontcolor", "light2",
"--bgcolor", "black",
])
run_gifmaker(command)
# This is an async example
async def run_gifmaker(command):
process = await asyncio.create_subprocess_shell(
command,
stdout=asyncio.subprocess.PIPE,
stderr=asyncio.subprocess.PIPE,
shell=True,
)
stdout, stderr = await process.communicate()
if process.returncode != 0:
print(f"Error: {stderr.decode()}")
return
await upload(Path(stdout.decode().strip()))