>On 17/01/2023 19:11, Paul B Mahol wrote:
>> On 1/17/23, Jeffrey Chapuis <ashyni1987 at gmail.com> wrote:
>>> I'm getting there, don't give up on me.
>>>
>>> Now 'limit_upscaled' become the variable used in filter_frame() and
>>> 'limit' is never change, unless by user
>>> We only have to update 'limit_upscaled' if 'limit' really changes.
>> 
>> probably ok
>
>>-    if (s->limit < 1.0)
>>-        s->limit *= (1 << desc->comp[0].depth) - 1;
>>+    s->bitdepth = desc->comp[0].depth;
>>+    s->limit_upscaled = s->limit;
>>+    if (s->limit_upscaled < 1.0)
>>+        s->limit_upscaled *= (1 << s->bitdepth) - 1;
>
>Is it better like this? we avoid a double assignment.
>
>-    if (s->limit < 1.0)
>-        s->limit *= (1 << desc->comp[0].depth) - 1;
>+    s->bitdepth = desc->comp[0].depth;
>+
>+    if (s->limit_upscaled < 1.0)
>+        s->limit_upscaled = s->limit * (1 << s->bitdepth) - 1;
>+    else
>+        s->limit_upscaled = s->limit;
>
>The full patch without this change is in the previous reply
>(https://ffmpeg.org/pipermail/ffmpeg-devel/2023-January/305806.html),
>let me know if there is other change to make with formatting/renaming.
>I really appreciate your patience with this patch.

Full patch attached without obvious errors from my previous proposal.