Add missing Pokémon sprites by anhthang · Pull Request #196 · PokeAPI/sprites

anhthang · 2026-02-14T15:47:33Z

This PR adds tooling and assets to help complete missing sprites:

Add a check_missing_sprites.py script that reads Pokémon and forms CSVs, determines generation via version groups, and reports/export CSVs of missing front/back (normal and shiny) sprites.
Add a smogon_download.py script to download missing sprites from the Smogon attachment URLs, then use renameSmogon.sh (updated to work from the scripts/downloads folder and move files instead of copy) to place them into the correct sprite directories.
Manually add sprites for special Imgur URLs and forms with pokemon_id ≥ 10000 that are not covered by the automated scripts.

Naramsim · 2026-02-16T08:35:09Z

Hi, thanks for the contribution! I currently don't have much time to review it, please give me some days.

anhthang · 2026-02-16T11:00:52Z

No worries at all, this PR turned out pretty big and I know it will take time to go through everything. Please take your time with the review.

I’ve tried to be careful with the Smogon links and file placements, so hopefully everything looks good there, but I’m happy to adjust anything you spot.

Naramsim · 2026-02-24T16:39:55Z

@anhthang The PR is too big to be reviewed, at least by me. Could you come up with a separate smaller one if possible (related to one script only maybe), so that it's going to be easier to review?

anhthang · 2026-02-25T02:51:53Z

@Naramsim This PR only depends on the download script and an update to the existing rename script to better handle forms/varieties. The check-missing script can be ignored for review purposes — it’s only used to identify missing assets and doesn’t affect the main workflow.

If splitting the PR would make it easier to review, which approach would you prefer? I could split the updates per generation sheet. However, Gen 9 might still be quite large since it includes many updates.

Another, though it may not be ideal, is reviewing by commit, since the changes are already logically separated across commits.

anhthang · 2026-02-28T04:31:23Z

@Naramsim I noticed that many images in the ‎⁠lowres⁠ folder don’t match the default sprite style, and the folder overall seems quite outdated.

Do we still need to keep the ‎⁠lowres⁠ folder around? If it’s no longer necessary, I’d be happy to open a separate PR to remove it safely.

Naramsim · 2026-02-28T09:08:22Z

@Naramsim I noticed that many images in the ‎⁠lowres⁠ folder don’t match the default sprite style, and the folder overall seems quite outdated.

Do we still need to keep the ‎⁠lowres⁠ folder around? If it’s no longer necessary, I’d be happy to open a separate PR to remove it safely.

Yes let's do it! First we remove the folder and then we merge this PR.

anhthang · 2026-02-28T09:55:13Z

Yes let's do it! First we remove the folder and then we merge this PR.

You wanna do it in this PR? My thought was that it might be better to do it later, once this PR and the other one for renaming varieties are in a good place.

Naramsim · 2026-03-01T17:19:19Z

Better in a separate PR

anhthang · 2026-03-02T02:05:30Z

Better in a separate PR

PR opened

…suffixes

scripts/missing_front_shiny.csv

scripts/pad_to_canvas.py

scripts/smogon_download.py

Naramsim · 2026-03-02T20:14:51Z

scripts/check_missing_sprites.py

+            # Export CSV for this category
+            file_safe_name = label.lower().replace(" ", "_")
+            pd.DataFrame(missing_list).to_csv(
+                f"missing_{file_safe_name}.csv", index=False


Does it make sense to run this script on our CICD let's say every week or month so to have a recap about which missing sprites are there?

Definitely. Probably once a month, or whenever there are changes in the sprites folder.

This is a cool idea.

This sounds like a pretty good idea we can schedule the action to run every time there is a merge to generate a report we can create a tracking issue which maintains a link to this report (maybe using this action https://github.com/JasonEtco/create-an-issue and use this field update_existing to update the issue everytime its run)

One point of concern is that the raw report might not fit in the issue body so alternatives would be

https://github.com/actions/upload-artifact uploading it as a workflow associated artefact

maybe using some filebin. gist etc (or we could also just write the report to one of the branches/folders in this repo)

Naramsim

Thanks for the job! Now I understand better the changes. @FallenDeity could you as well take a look here at the files check_missing_sprites.py, smogon_download.py and renameSmogon.sh?

anhthang · 2026-03-03T05:13:55Z

I've added documentation for the new scripts so it's easier to understand and maintain the workflow going forward.

Once this PR is merged, issues #76 and #104 can be considered completed. The remaining missing sprites can be updated later once Smogon finishes adding them on their side.

scripts/check_missing_sprites.py

FallenDeity · 2026-03-03T13:27:01Z

scripts/forms.json

+    "896": "glastrier",
+    "897": "spectrier",
+    "978": "tatsugiri-curly",
+    "658_2": "greninja-battle-bond",


is this intended the order is not maintained anymore?

I don’t think the old script was actually maintaining the order either, but I can revert it to minimize changes and make the review easier.

FallenDeity · 2026-03-03T13:32:54Z

scripts/missing_sprites.csv

As @Naramsim raised I think it would be better to have this record csv as a part of the pipeline along with maybe a summary report too with categorization of sprites broken down like no. missing, and some diffs metrics and breakdown between back, front, shinies etc. The artifact upload step stores it as a zip so we could have some detailed records stored there

I haven’t looked into the pipeline yet, but I think the current report is good enough for now. We can definitely extend it later to include a CSV artifact and a more detailed breakdown once we align on the exact format we want.

Yeah i can write the pipeline after we merge this PR

sounds good

scripts/showdown.py

scripts/check_missing_sprites.py

FallenDeity · 2026-03-03T14:12:23Z

scripts/renameSmogon.sh

    local pokemonName
    local inputGeneration="${1:-6}"
-    local files="../smogon/gen$inputGeneration/"
+    local files="downloads/"


Why is it downloads now? we can just keep it smogon gen specific right, I think a little comment regarding files source img dir and how pokeapi sprites (i.e destination root relates to that in terms of placing directories).

Currently it attempts to go two directories up from downloads? the behavior is not clear

It’s just the folder path/name used to store images downloaded from Smogon — there’s no logic change here. I only updated it to use the downloads directory, which is where smogon_download.py saves the images.

alright sounds good although I prefer stuff to happen relative to my root project dir but downloads work here if the download script uses it I suppose

FallenDeity · 2026-03-03T14:14:41Z

scripts/renameSmogon.sh

                else
                    echo "[+] Copying GMax $smogonName to $destination/$pokemonID.png"
-                    cp "$smogonName" "$destination/$pokemonID.png"
+                    mv "$smogonName" "$destination/$pokemonID.png"


mv consumes the original file so if the next if statement is triggered after this it will fail since mv will no longer find the original file, which will crash the script

FallenDeity · 2026-03-03T14:30:15Z

scripts/renameSmogon.sh

+                        # --- FALLBACK LOGIC: Search 'forms' using the full name ---
+                        echo "[!] $pokemonName not found directly. Searching forms for Species ID: $id..."
+
+                        speciesResponse=$(curl -sS "https://pokeapi.co/api/v2/pokemon/$id/" 2>/dev/null)


Previously the $id is being used to hit the species endpoint in the gmax if and here its in pokemon endpoint is that intended or can we directly hit the species endpoint because pokemon forms and species varieties give different results for example pikachu (25)

I'm using the Pokémon endpoint here to properly cover different varieties (e.g. Pikachu caps, Vivillon forms, etc.). Those can’t reliably be resolved from the species endpoint by name alone, since species and forms/varieties return different structures and data.

So this change ensures we correctly handle form-specific cases rather than just the base species.

but the pokemon endpoint for say pikachu dosent give all the cap variants the species endpoint does that if you check the forms field its just pikachu there

The old logic is still here (https://github.com/PokeAPI/sprites/pull/196/changes#diff-2074368bbccd08fa9ceeafb33f6af9f1f93044835d3cd67eb74eb4702b257cccR93). If the variants aren’t available from the Pokémon endpoint, we’ll fall back to the species endpoint to retrieve them.

you are right that the old logic is still there (it hits the pokemon endpoint to check)

but in your fallback logic you are again hitting the pokemon endpoint itself did you mean to hit the pokemon-species endpoint?

# --- FALLBACK LOGIC: Search 'forms' using the full name --- echo "[!] $pokemonName not found directly. Searching forms for Species ID: $id..." speciesResponse=$(curl -sS "https://pokeapi.co/api/v2/pokemon/$id/" 2>/dev/null)

because for pokemon endpoint for pikachu https://pokeapi.co/api/v2/pokemon/25

"forms": [ { "name": "pikachu", "url": "https://pokeapi.co/api/v2/pokemon-form/25/" } ],

species endpoint https://pokeapi.co/api/v2/pokemon-species/25/

"varieties": [ { "is_default": true, "pokemon": { "name": "pikachu", "url": "https://pokeapi.co/api/v2/pokemon/25/" } }, { "is_default": false, "pokemon": { "name": "pikachu-rock-star", "url": "https://pokeapi.co/api/v2/pokemon/10080/" } }, { "is_default": false, "pokemon": { "name": "pikachu-belle", "url": "https://pokeapi.co/api/v2/pokemon/10081/" } }, ...

or it might be smart to cover both forms and varieties I am not entirely sure how this distinction is draw in pokeapi in bulbapedia they just list everything as form diffs https://bulbapedia.bulbagarden.net/wiki/List_of_Pok%C3%A9mon_with_form_differences

but in pokeapi we have pokemon -> forms and followed by species -> varieties I think we should cover both in this fallback logic? @Naramsim what are your thoughts on this?

Yes, it's an old debate. Nobody know now the true difference between forms, variants, species.

FallenDeity · 2026-03-03T14:36:06Z

scripts/smogon_download.py

+import requests
+import re
+
+# The list of URLs you provided


Add the google doc link url here or a pointer to the forum/discussion? its not clear where the urls are coming from, if its a public google sheet cant we just download it using requests etc and parse that instead of hardcoding urls? this script feels a bit redundant

It's fine to explain again here (I mentioned it in the doc too).

The URLs are image links from the Smogon spreadsheet. The issue is that some sprite links have wrong filenames, duplicate IDs, and a few are Imgur links that don’t follow the same naming format. Because of that, we can’t reliably auto-download and rename them with a consistent rule.

Since the content needs manual verification anyway, I added the links explicitly to avoid broken mappings.

But once a download is done by the pr and merged those urls are no longer relevant right also can you share the spreadsheet url here I wasn't able to find it 😅

I already added the Smogon thread in the latest commit — you’ll find all the spreadsheet links in that topic.

FallenDeity · 2026-03-03T14:52:12Z

Also just another note some images just seem to be updated with no diffs at all https://github.com/PokeAPI/sprites/pull/196/changes#diff-fbf2451a323761b6e6ee4b4f289800136c2a33650eaa9c279ff54301bb13ed2c

Is this happening because the script is overwriting existing files during the form logic? if there is such a case maybe compare the image contents and if there is a diff log it and replace, if not just delete the <id>-<form_suffix>.png this way there will only be a deletion diff and not both a delete + replacement (the diffs wont be as noisy)

anhthang · 2026-03-03T16:04:06Z

Also just another note some images just seem to be updated with no diffs at all https://github.com/PokeAPI/sprites/pull/196/changes#diff-fbf2451a323761b6e6ee4b4f289800136c2a33650eaa9c279ff54301bb13ed2c

Is this happening because the script is overwriting existing files during the form logic? if there is such a case maybe compare the image contents and if there is a diff log it and replace, if not just delete the <id>-<form_suffix>.png this way there will only be a deletion diff and not both a delete + replacement (the diffs wont be as noisy)

You're right — that's essentially what’s happening.

For a few images I’ve already noticed there are slight differences (just a few pixels changed), but I don’t remember all of them off the top of my head.

…plaination

…kemon id

FallenDeity

Reviewed just two minor changes rest of the stuff looks good

FallenDeity · 2026-03-05T13:04:15Z

scripts/missing_sprites.csv

sounds good

FallenDeity · 2026-03-05T13:05:49Z

scripts/renameSmogon.sh

    local pokemonName
    local inputGeneration="${1:-6}"
-    local files="../smogon/gen$inputGeneration/"
+    local files="downloads/"


alright sounds good although I prefer stuff to happen relative to my root project dir but downloads work here if the download script uses it I suppose

FallenDeity · 2026-03-05T13:17:31Z

scripts/renameSmogon.sh

+                        # --- FALLBACK LOGIC: Search 'forms' using the full name ---
+                        echo "[!] $pokemonName not found directly. Searching forms for Species ID: $id..."
+
+                        speciesResponse=$(curl -sS "https://pokeapi.co/api/v2/pokemon/$id/" 2>/dev/null)


you are right that the old logic is still there (it hits the pokemon endpoint to check)

but in your fallback logic you are again hitting the pokemon endpoint itself did you mean to hit the pokemon-species endpoint?

# --- FALLBACK LOGIC: Search 'forms' using the full name --- echo "[!] $pokemonName not found directly. Searching forms for Species ID: $id..." speciesResponse=$(curl -sS "https://pokeapi.co/api/v2/pokemon/$id/" 2>/dev/null)

because for pokemon endpoint for pikachu https://pokeapi.co/api/v2/pokemon/25

"forms": [ { "name": "pikachu", "url": "https://pokeapi.co/api/v2/pokemon-form/25/" } ],

species endpoint https://pokeapi.co/api/v2/pokemon-species/25/

"varieties": [ { "is_default": true, "pokemon": { "name": "pikachu", "url": "https://pokeapi.co/api/v2/pokemon/25/" } }, { "is_default": false, "pokemon": { "name": "pikachu-rock-star", "url": "https://pokeapi.co/api/v2/pokemon/10080/" } }, { "is_default": false, "pokemon": { "name": "pikachu-belle", "url": "https://pokeapi.co/api/v2/pokemon/10081/" } }, ...

or it might be smart to cover both forms and varieties I am not entirely sure how this distinction is draw in pokeapi in bulbapedia they just list everything as form diffs https://bulbapedia.bulbagarden.net/wiki/List_of_Pok%C3%A9mon_with_form_differences

but in pokeapi we have pokemon -> forms and followed by species -> varieties I think we should cover both in this fallback logic? @Naramsim what are your thoughts on this?

FallenDeity · 2026-03-05T13:44:09Z

scripts/sort_forms.py

I don't think this is required can be simplified to a jq command like this which we can put in README, or run in the pipeline if required

jq 'to_entries | sort_by( .key | (capture("^(?<n>\\d+)(?<r>.*)") // {n: "999999999", r: .}) | (.n | tonumber) as $m | .r | if . == "" then [$m, 0, 0, ""] elif test("^_\\d+$") then [$m, 1, (.[1:] | tonumber), ""] elif test("^[a-zA-Z]+$") then [$m, 2, 0, .] elif startswith("_") then [$m, 3, 0, .] else [$m, 4, 0, .] end ) | from_entries' forms.json > sorted_forms.json

thanks. run in pipeline is good idea and @Naramsim can do it I believe

anhthang · 2026-03-05T13:55:29Z

@FallenDeity Thanks for pointing that out about the rename script. The first call is by name and the fallback is by dex number on the same endpoint. In my mind the fallback should use ‎⁠pokemon-species⁠ to pull varieties, but it seems to work. Let me double-check that logic again.

Edit: pokemonName comes from the pokemon-species endpoint.

To determine the correct Pokémon ID, we first attempt to fetch the Pokémon using the variety-name. If it exists, we use the returned Pokémon ID.
If it does not exist, we fetch the Pokémon’s forms using the Pokémon ID and check whether the form-name matches. When a matching form is found, the image is renamed using the format <id>-<form-name>

Naramsim · 2026-03-06T16:41:30Z

Alright, @anhthang is this mergeable? I don't know if there's more work to do.

anhthang · 2026-03-06T22:18:54Z

Yes please merge the PR @Naramsim, then I will sync this change to generation-v folder and do rename varieties for that

anhthang force-pushed the add-missing-sprites branch from 6bb59fb to d864acc Compare February 20, 2026 04:49

anhthang mentioned this pull request Mar 2, 2026

Remove lowres Pokémon sprites #200

Merged

anhthang added 19 commits March 2, 2026 14:52

add script to check missing sprites

f6b3d48

add missing front, front shiny sprites

e3a2ae7

add gen 6 back

24c4a7d

add missing gen 8 sprites

28138c5

add missing gen 6 shiny back

d0019b1

ruff format

f83c0e6

fix: optimize sprite renaming with species-form fallback and cleaner …

7803221

…suffixes

fixing silvally sprites

bb09490

add missing alcremie sprites

6fe29f5

add missing pikachu region cap

c88fcba

final addition

f3e45df

fix: front sceptile mega

a10cc00

rerun gen9 and update all sprites from latest

e51586f

fix Xerneas

1b86cac

rename to correct pokemon id, remove some duplicate images

491b447

rename part 2

cc10418

fix xerneas shiny

1b2dc22

revert ampharos mega

9dbc174

fix Zarude-dada back

8a2886d

anhthang added 2 commits March 2, 2026 14:52

fix shiny rotom

7dd1eb1

fix: rapidash, zorua color update

98b7acb

anhthang force-pushed the add-missing-sprites branch from 31540a8 to 98b7acb Compare March 2, 2026 09:14

Naramsim reviewed Mar 2, 2026

View reviewed changes

scripts/missing_front_shiny.csv Outdated Show resolved Hide resolved

Naramsim reviewed Mar 2, 2026

View reviewed changes

scripts/pad_to_canvas.py Show resolved Hide resolved

Naramsim reviewed Mar 2, 2026

View reviewed changes

scripts/smogon_download.py Outdated Show resolved Hide resolved

Naramsim reviewed Mar 2, 2026

View reviewed changes

anhthang mentioned this pull request Mar 3, 2026

Rename variety sprites: Gen IV/VI #202

Merged

anhthang added 2 commits March 3, 2026 10:55

chore: add contributing document

9880d09

fix: consolidate missing sprite checks into single report

3b7ab05

Naramsim approved these changes Mar 3, 2026

View reviewed changes

FallenDeity reviewed Mar 3, 2026

View reviewed changes

FallenDeity suggested changes Mar 3, 2026

View reviewed changes

anhthang requested a review from FallenDeity March 3, 2026 16:28

fix: use pathlib, get csv data from upstream, add more comment for ex…

08039c8

…plaination

anhthang force-pushed the add-missing-sprites branch from 6f923b5 to 08039c8 Compare March 3, 2026 16:31

anhthang added 4 commits March 4, 2026 16:59

add sort forms to make it consistent later

e9ad275

fix: do not copy if no suffix found to prevent overritting default po…

0692132

…kemon id

fix: palafin

4a000df

revert to ogerpon mask

242b4b9

FallenDeity approved these changes Mar 5, 2026

View reviewed changes

fix: clarify the logic for renaming to id-formname format

3e63869

Conversation

anhthang commented Feb 14, 2026

Uh oh!

Naramsim commented Feb 16, 2026

Uh oh!

anhthang commented Feb 16, 2026

Uh oh!

Naramsim commented Feb 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

anhthang commented Feb 25, 2026

Uh oh!

anhthang commented Feb 28, 2026

Uh oh!

Naramsim commented Feb 28, 2026

Uh oh!

anhthang commented Feb 28, 2026

Uh oh!

Naramsim commented Mar 1, 2026

Uh oh!

anhthang commented Mar 2, 2026

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Naramsim left a comment

Choose a reason for hiding this comment

Uh oh!

anhthang commented Mar 3, 2026

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Naramsim commented Feb 24, 2026 •

edited

Loading

FallenDeity commented Mar 3, 2026 •

edited

Loading

anhthang commented Mar 5, 2026 •

edited

Loading

Naramsim commented Mar 6, 2026 •

edited

Loading