To decode these, we throw away the last tx and the last constant of each input, e.g.:

btc getrawtransaction 033d185d1a04c4bd6de9bb23985f8c15aa46234206ad29101c31f4b33f1a0e49 true | jq -r '.vin[].scriptSig.asm' | head -n -1 | sed -r 's/ [^ ]+$//' | tr -d '\n'  | xxd -r -p > tmp.jpg

Terminology mentioned e.g. at: Data Insertion in Bitcoin's Blockchain by Andrew Sward, Vecna OP_0 and Forrest Stonedahl.

Daisy chain Bitcoin inscription

 0  0 

This is a term invented by Ciro Santilli, and refers to a loose set of uncommon Bitcoin inscription methods that involve inscribing one or a small number of payloads per Bitcoin transaction.

These methods are both inefficient and hard to detect and decode, partly because Bitcoin Core does not index spending transactions: bitcoin.stackexchange.com/questions/61794/bitcoin-rpc-how-to-find-the-transaction-that-spends-a-txo. This makes finding them all that more rewarding however.

On the other hand, they do have the advantage of not depending on any block size limits, as their individual transactions are very small.

Inscribing anything large would however take a very long time, as you'd have to wait until the previous payload chunk is confirmed before going to the next one. This alone makes the format impractical perhaps.

Known examples: