diff --git a/docs/common_use_cases.rst b/docs/common_use_cases.rst
index 2fb1e69192..e9af89fb22 100644
--- a/docs/common_use_cases.rst
+++ b/docs/common_use_cases.rst
@@ -120,9 +120,9 @@ Some tips may help you to get better results.
 - Tables are known to be slow, especially when they are rendered on multiple
   pages. When possible, using a common block layout instead gives much faster
   layouts.
-- Encoding detection can be really slow when HTML lines are really long.
-  Providing an explicit encoding or removing the ``chardet`` module fixes the
-  problem (see `#29`_).
+- Optimizing images and fonts can reduce the PDF size, but increase the
+  rendering time. Moreover, caching images gives the possibility to read and
+  optimize images only once, and thus to save time when the same image is used
+  multiple times. See :ref:`Image Cache and Optimization`.
 
 .. _WeasyPerf: https://kozea.github.io/WeasyPerf/
-.. _#29: https://github.com/chardet/chardet/issues/29
diff --git a/docs/first_steps.rst b/docs/first_steps.rst
index d5b0fb5640..9f9b5e1423 100644
--- a/docs/first_steps.rst
+++ b/docs/first_steps.rst
@@ -507,6 +507,41 @@ the function internally used by WeasyPrint to retreive data.
 .. _Django-WeasyPrint: https://github.com/fdemmer/django-weasyprint
 .. _Django: https://www.djangoproject.com/
 
+Image Cache and Optimization
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+WeasyPrint provides two options to deal with images: ``optimize_size`` and
+``image_cache``.
+
+``optimize_size`` can enable size optimization for images, but also for fonts.
+When enabled, the generated PDF will include smaller images and fonts, but the
+rendering time may be slightly increased.
+
+.. code-block:: python
+
+    # No size optimization, faster, but generated PDF is larger
+    HTML('http://example.org/').write_pdf(
+        'example.pdf', optimize_size=())
+
+    # Full size optimization, slower, but generated PDF is smaller
+    HTML('http://example.org/').write_pdf(
+        'example.pdf', optimize_size=('fonts', 'images'))
+
+``image_cache`` gives the possibility to use a cache for images, avoiding to
+download, parse and optimize them each time they are used.
+
+By default, the cache is used document by document, but you can share it
+between documents if needed. This feature can save a lot of network and CPU
+time when you render a lot of documents that use the same images.
+
+.. code-block:: python
+
+    cache = {}
+    for i in range(10):
+        HTML(f'http://example.org/?id={i}').write_pdf(
+            f'example-{i}.pdf', image_cache=cache)
+
+
 Logging
 ~~~~~~~