[Ffmpeg-devel] [PATCH] flash screen video encoder

[Benjamin Larsson]

Michael Niedermayer wrote:

>Hi
>
>On Sun, Jan 21, 2007 at 05:39:32PM +0100, Benjamin Larsson wrote:
>  
>
>>Hi, thanks for the review.
>>
>>The plan is to get a working encoder commited and then later add 2 pass
>>and optimizations.
>>
>>
>>    
>>
>
>[...]
>  
>
>>>[...]
>>> 
>>>
>>>      
>>>
>>>>+                ptr[0] = 0;
>>>>+                ptr[1] = 0;
>>>>+                buf_pos +=2;
>>>>   
>>>>
>>>>        
>>>>
>>>please use bytestream.h
>>> 
>>>
>>>      
>>>
>>Do you mean bytestream_put_be16(ptr,0) ?
>>    
>>
>
>yes exactly
>
>
>  
>
>>> 
>>>
>>>      
>>>
>>>>+            }
>>>>+        }
>>>>+    }
>>>>+
>>>>+    if (pred_blocks)
>>>>+        *I_frame = 0;
>>>>+    else
>>>>+        *I_frame = 1;
>>>>   
>>>>
>>>>        
>>>>
>>>actually you dont need to keep track of pred_blocks, a simple check for the
>>>buf_size vs. the number of blocks*2 should do i think
>>> 
>>>
>>>      
>>>
>>I prefer it this way as I could extend the I frame decision then. Think
>>of a frame that all but one block changes. Then it might be of advantage
>>to encode it as an I frame. The format only allow block changes at I
>>frames.
>>    
>>
>
>ok
>
>[...]
>
>
>
>[...]
>  
>
>>+static int flashsv_encode_init(AVCodecContext *avctx)
>>+{
>>+    FlashSVContext *s = (FlashSVContext *)avctx->priv_data;
>>+
>>+    s->avctx = avctx;
>>+
>>+    if ((avctx->width > 4095) || (avctx->height > 4095)) {
>>+        av_log(avctx, AV_LOG_ERROR, "Input dimensions to large, input must be max 4096x4096 !\n");
>>+        return -1;
>>+    }
>>+
>>+    if (avcodec_check_dimensions(avctx, avctx->width, avctx->height) < 0) {
>>+        return -1;
>>+    }
>>+
>>+    s->first_frame = 1;
>>+
>>+    // Needed if zlib unused or init aborted before deflateInit
>>+    memset(&(s->zstream), 0, sizeof(z_stream));
>>+/*
>>+    s->zstream.zalloc = NULL; //av_malloc;
>>+    s->zstream.zfree = NULL; //av_free;
>>+    s->zstream.opaque = NULL;
>>+    zret = deflateInit(&(s->zstream), 9);
>>+    if (zret != Z_OK) {
>>+        av_log(avctx, AV_LOG_ERROR, "Inflate init error: %d\n", zret);
>>+        return -1;
>>+    }
>>+*/
>>+
>>+    s->image_width = avctx->width;
>>+    s->image_height = avctx->height;
>>+
>>+    if ((s->tmpblock = av_mallocz(3*256*256)) == NULL) {
>>+        av_log(avctx, AV_LOG_ERROR, "Can't allocate compression buffer.\n");
>>+        return -1;
>>+    }
>>+
>>+    if (!s->encbuffer)
>>+        s->encbuffer = av_mallocz(s->image_width*s->image_height*3);
>>    
>>
>
>how can the buffer be already allocated? and if so how can we be sure its
>old size is large enough?
>  
>

Fixed.

>
>  
>
>>+
>>+
>>+    return 0;
>>+}
>>+
>>+#ifdef CONFIG_ENCODERS
>>    
>>
>
>the 2 functions above are also encoder specific i think and it would be nice
>if the encoder and decoder would be in seperate files, would avoid all the
>CONFIG_EN/DECODERS
>
>[...]
>
>  
>

Fixed.

And new patch attached.

MvH
Benjamin Larsson

-- 
new tiny signature

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: flashsv3.patch
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20070121/69879c9d/attachment.asc>