mirror/LocalAI
1
0
Fork 0
mirror of https://github.com/mudler/LocalAI.git synced 2026-06-09 01:07:09 -04:00
Code Issues Packages Projects Releases Wiki Activity

feat(llama-cpp): video input support (mtmd #24269) (#10216)

Browse Source 
* chore(llama-cpp): bump to 8f83d6c for mtmd video input support

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

* feat(llama-cpp): forward video input to mtmd (template + non-template paths)

Wire request->videos() into grpc-server.cpp mirroring the existing image
and audio handling: a video_data build + non-template files extraction, and
input_video chat chunks on the tokenizer-template path. allow_video is
auto-set at model load by the vendored upstream chat_params.

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

* feat(ui): add video attachment support to the chat UI

Mirror the image/audio attachment path for video: emit video_url content
parts, accept video/* in the picker, keep video files as base64, show a
film icon badge, and render attached video inline with a <video> player.

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

* fix(llama-cpp): patch mtmd video stdin double-close (heap crash)

Upstream mtmd video input (ggml-org/llama.cpp#24269) double-fcloses the
ffmpeg/ffprobe stdin FILE: feed_stdin() fclose()s the FILE returned by
subprocess_stdin() (which is sp->stdin_file), then subprocess_destroy()
fclose()s the same pointer again -> heap corruption that aborts the
backend on any base64 input_video request (the CLI --video file path is
unaffected). Vendor a one-line fix (null sp->stdin_file after fclose)
via prepare.sh's patches/ until upstream merges it.

Verified e2e with gemma-4-e2b-it-qat-q4_0: video frames decode via
ffmpeg and the model answers correctly (red clip -> 'Red', blue -> 'Blue').

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

* chore(llama-cpp): re-pin to upstream #24316, drop vendored stdin patch

Upstream replaced the ad-hoc video stdin handling with a proper RAII
refactor (ggml-org/llama.cpp#24316, "mtmd: refactor video subproc
handling"), which includes the same `sp->stdin_file = nullptr` guard our
patch added (plus join-before-destroy ordering). Re-pin LLAMA_VERSION to
that branch head and drop patches/0001 - it's now redundant.

Verified e2e with gemma-4-e2b-it-qat-q4_0: no crash, video frames decode
and the model answers correctly (red clip -> "Red", blue -> "Blue").

NOTE: #24316 is not yet merged, so this pins to its branch-head commit
(28ca1e60). Re-pin to the squash-merge commit on master once it lands,
otherwise `git fetch` may lose the commit after the branch is deleted.

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

---------

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
Co-authored-by: Ettore Di Giacinto <mudler@localai.io>
This commit is contained in:
LocalAI [bot]
2026-06-08 23:17:50 +02:00
committed by GitHub
parent c20225fc13
commit 9323f4b5ca
4 changed files with 85 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
backend/cpp/llama-cpp/Makefile
View File

@@ -1,5 +1,5 @@
LLAMA_VERSION?=9e3b928fd8c9d14dbf15a8768b9fdd7e5c721d66
LLAMA_VERSION?=28ca1e600c5dac1854fb7e09611914013430b037
LLAMA_REPO?=https://github.com/ggerganov/llama.cpp
CMAKE_ARGS?=

73
backend/cpp/llama-cpp/grpc-server.cpp
View File

@@ -381,6 +381,15 @@ json parse_options(bool streaming, const backend::PredictOptions* predict, const
});
}
// for each video in the request, add the video data
for (int i = 0; i < predict->videos_size(); i++) {
data["video_data"].push_back(json
{
{"id", i},
{"data", predict->videos(i)},
});
}
data["stop"] = predict->stopprompts();
// data["n_probs"] = predict->nprobs();
//TODO: images,
@@ -1503,7 +1512,7 @@ public:
msg_json["role"] = msg.role();
bool is_last_user_msg = (i == last_user_msg_idx);
bool has_images_or_audio = (request->images_size() > 0 || request->audios_size() > 0);
bool has_images_or_audio = (request->images_size() > 0 || request->audios_size() > 0 || request->videos_size() > 0);
// Handle content - can be string, null, or array
// For multimodal content, we'll embed images/audio from separate fields
@@ -1554,6 +1563,16 @@ public:
content_array.push_back(audio_chunk);
}
}
if (request->videos_size() > 0) {
for (int j = 0; j < request->videos_size(); j++) {
json video_chunk;
video_chunk["type"] = "input_video";
json input_video;
input_video["data"] = request->videos(j);
video_chunk["input_video"] = input_video;
content_array.push_back(video_chunk);
}
}
msg_json["content"] = content_array;
} else {
// Use content as-is (already array or not last user message)
@@ -1588,6 +1607,16 @@ public:
content_array.push_back(audio_chunk);
}
}
if (request->videos_size() > 0) {
for (int j = 0; j < request->videos_size(); j++) {
json video_chunk;
video_chunk["type"] = "input_video";
json input_video;
input_video["data"] = request->videos(j);
video_chunk["input_video"] = input_video;
content_array.push_back(video_chunk);
}
}
msg_json["content"] = content_array;
} else if (msg.role() == "tool") {
// Tool role messages must have content field set, even if empty
@@ -2039,6 +2068,16 @@ public:
files.push_back(decoded_data);
}
}
const auto &video_data = data.find("video_data");
if (video_data != data.end() && video_data->is_array())
{
for (const auto &video : *video_data)
{
auto decoded_data = base64_decode(video["data"].get<std::string>());
files.push_back(decoded_data);
}
}
}
const bool has_mtmd = ctx_server.impl->mctx != nullptr;
@@ -2291,7 +2330,7 @@ public:
}
bool is_last_user_msg = (i == last_user_msg_idx);
bool has_images_or_audio = (request->images_size() > 0 || request->audios_size() > 0);
bool has_images_or_audio = (request->images_size() > 0 || request->audios_size() > 0 || request->videos_size() > 0);
// Handle content - can be string, null, or array
// For multimodal content, we'll embed images/audio from separate fields
@@ -2344,6 +2383,16 @@ public:
content_array.push_back(audio_chunk);
}
}
if (request->videos_size() > 0) {
for (int j = 0; j < request->videos_size(); j++) {
json video_chunk;
video_chunk["type"] = "input_video";
json input_video;
input_video["data"] = request->videos(j);
video_chunk["input_video"] = input_video;
content_array.push_back(video_chunk);
}
}
msg_json["content"] = content_array;
} else {
// Use content as-is (already array or not last user message)
@@ -2383,6 +2432,16 @@ public:
content_array.push_back(audio_chunk);
}
}
if (request->videos_size() > 0) {
for (int j = 0; j < request->videos_size(); j++) {
json video_chunk;
video_chunk["type"] = "input_video";
json input_video;
input_video["data"] = request->videos(j);
video_chunk["input_video"] = input_video;
content_array.push_back(video_chunk);
}
}
msg_json["content"] = content_array;
SRV_INF("[CONTENT DEBUG] Predict: Message %d created content array with media\n", i);
} else if (!msg.tool_calls().empty()) {
@@ -2845,6 +2904,16 @@ public:
files.push_back(decoded_data);
}
}
const auto &video_data = data.find("video_data");
if (video_data != data.end() && video_data->is_array())
{
for (const auto &video : *video_data)
{
auto decoded_data = base64_decode(video["data"].get<std::string>());
files.push_back(decoded_data);
}
}
}
// process files

6
core/http/react-ui/src/hooks/useChat.js vendored
View File

@@ -216,6 +216,12 @@ export function useChat(initialModel = '') {
audio_url: { url: `data:${file.type};base64,${file.base64}` },
})
userFiles.push({ name: file.name, type: 'audio' })
} else if (file.type?.startsWith('video/')) {
messageContent.push({
type: 'video_url',
video_url: { url: `data:${file.type};base64,${file.base64}` },
})
userFiles.push({ name: file.name, type: 'video' })
} else {
// Text/PDF files - append to content
if (file.textContent) {

11
core/http/react-ui/src/pages/Chat.jsx
View File

@@ -265,7 +265,7 @@ function UserMessageContent({ content, files }) {
<div className="chat-message-files">
{files.map((f, i) => (
<span key={i} className="chat-file-inline">
<i className={`fas ${f.type === 'image' ? 'fa-image' : f.type === 'audio' ? 'fa-headphones' : 'fa-file'}`} />
<i className={`fas ${f.type === 'image' ? 'fa-image' : f.type === 'audio' ? 'fa-headphones' : f.type === 'video' ? 'fa-film' : 'fa-file'}`} />
{f.name}
</span>
))}
@@ -274,6 +274,9 @@ function UserMessageContent({ content, files }) {
{Array.isArray(content) && content.filter(c => c.type === 'image_url').map((img, i) => (
<img key={i} src={img.image_url.url} alt="attached" className="chat-inline-image" />
))}
{Array.isArray(content) && content.filter(c => c.type === 'video_url').map((vid, i) => (
<video key={i} src={vid.video_url.url} controls className="chat-inline-video" />
))}
</>
)
}
@@ -711,7 +714,7 @@ export default function Chat() {
for (const file of e.target.files) {
const base64 = await fileToBase64(file)
const entry = { name: file.name, type: file.type, base64 }
if (!file.type.startsWith('image/') && !file.type.startsWith('audio/')) {
if (!file.type.startsWith('image/') && !file.type.startsWith('audio/') && !file.type.startsWith('video/')) {
entry.textContent = await file.text().catch(() => '')
}
newFiles.push(entry)
@@ -1244,7 +1247,7 @@ export default function Chat() {
<div className="chat-files">
{files.map((f, i) => (
<span key={i} className="chat-file-badge">
<i className={`fas ${f.type?.startsWith('image/') ? 'fa-image' : f.type?.startsWith('audio/') ? 'fa-headphones' : 'fa-file'}`} />
<i className={`fas ${f.type?.startsWith('image/') ? 'fa-image' : f.type?.startsWith('audio/') ? 'fa-headphones' : f.type?.startsWith('video/') ? 'fa-film' : 'fa-file'}`} />
{f.name}
<button onClick={() => setFiles(prev => prev.filter((_, idx) => idx !== i))}>
<i className="fas fa-xmark" />
@@ -1343,7 +1346,7 @@ export default function Chat() {
ref={fileInputRef}
type="file"
multiple
accept="image/*,audio/*,application/pdf,.txt,.md,.csv,.json"
accept="image/*,audio/*,video/*,application/pdf,.txt,.md,.csv,.json"
style={{ display: 'none' }}
onChange={handleFileChange}
/>